diff --git "a/questions/MLS-C01-v0624.json" "b/questions/MLS-C01-v0624.json" new file mode 100644--- /dev/null +++ "b/questions/MLS-C01-v0624.json" @@ -0,0 +1,3182 @@ +[ + { + "question": "A Machine Learning Specialist is working with multi ple data sources containing billions of records that need to be joined. What feature engineering an d model development approach should the Specialist take with a dataset this large?", + "options": [ + "A. Use an Amazon SageMaker notebook for both feature engineering and model development", + "B. Use an Amazon SageMaker notebook for feature engi neering and Amazon ML for model development", + "C. Use Amazon EMR for feature engineering and Amazon SageMaker SDK for model development", + "D. Use Amazon ML for both feature engineering and mo del development." + ], + "correct": "C. Use Amazon EMR for feature engineering and Amazon SageMaker SDK for model development", + "explanation": "Amazon EMR is a service that can process large amou nts of data efficiently and cost-effectively. It can run distributed frameworks such as Apache Spark , which can perform feature engineering on big data. Amazon SageMaker SDK is a Python library that can interact with Amazon SageMaker service to train and deploy machine learning models. It can al so use Amazon EMR as a data source for training data. Reference: Amazon EMR Amazon SageMaker SDK", + "references": "" + }, + { + "question": "A Machine Learning Specialist has completed a proof of concept for a company using a small data sample and now the Specialist is ready to implement an end-to-end solution in AWS using Amazon SageMaker The historical training data is stored in Amazon RDS Which approach should the Specialist use for traini ng a model using that data?", + "options": [ + "A. Write a direct connection to the SQL database wit hin the notebook and pull data in", + "B. Push the data from Microsoft SQL Server to Amazon S3 using an AWS Data Pipeline and provide the S3", + "C. Move the data to Amazon DynamoDB and set up a con nection to DynamoDB within the notebook to pull", + "D. Move the data to Amazon ElastiCache using AWS DMS and set up a connection within the notebook to pul l", + "A. Recall", + "B. Misclassification rate", + "C. Mean absolute percentage error (MAPE)", + "D. Area Under the ROC Curve (AUC)" + ], + "correct": "D. Area Under the ROC Curve (AUC)", + "explanation": "Area Under the ROC Curve (AUC) is a metric that mea sures the performance of a binary classifier across all possible thresholds. It is also known as the probability that a randomly chosen positive example will be ranked higher than a randomly chose n negative example by the classifier. AUC is a good metric to compare different classification mod els because it is independent of the class distribution and the decision threshold. It also ca ptures both the sensitivity (true positive rate) an d the specificity (true negative rate) of the model. Reference: AWS Machine Learning Specialty Exam Guide AWS Machine Learning Specialty Sample Questions", + "references": "Amazon SageMaker AWS Data Pipeline QUESTION 3 Which of the following metrics should a Machine Lea rning Specialist generally use to compare/evaluate machine learning classification mo dels against each other?" + }, + { + "question": "A Machine Learning Specialist is using Amazon Sage Maker to host a model for a highly available customer-facing application. The Specialist has trained a new version of the mod el, validated it with historical data, and now wants to deploy it to production To limit any risk of a negative customer experience, the Specialist wants to be able to monitor the model and roll it b ack, if needed What is the SIMPLEST approach with the LEAST risk t o deploy the model and roll it back, if needed?", + "options": [ + "A. Create a SageMaker endpoint and configuration for the new model version. Redirect production", + "B. Create a SageMaker endpoint and configuration for the new model version. Redirect production", + "C. Update the existing SageMaker endpoint to use a n ew configuration that is weighted to send 5%", + "D. Update the existing SageMaker endpoint to use a n ew configuration that is weighted to send" + ], + "correct": "C. Update the existing SageMaker endpoint to use a n ew configuration that is weighted to send 5%", + "explanation": "Updating the existing SageMaker endpoint to use a n ew configuration that is weighted to send 5% of the traffic to the new variant is the simplest appr oach with the least risk to deploy the model and ro ll it back, if needed. This is because SageMaker suppo rts A/B testing, which allows the Specialist to compare the performance of different model variants by sending a portion of the traffic to each variant. The Specialist can monitor the metrics of each variant and adjust the weights accordingly. If the new variant does not perform as expected, the S pecialist can revert traffic to the last version by resetting the weights to 100% for the old variant a nd 0% for the new variant. This way, the Specialistcan deploy the model without affecting the customer experience and roll it back easily if needed. Reference: Amazon SageMaker Deploying models to Amazon SageMaker hosting servic es", + "references": "" + }, + { + "question": "A manufacturing company has a large set of labeled historical sales data The manufacturer would like to predict how many units of a particular part shou ld be produced each quarter Which machine learning approach should be used to solve this prob lem?", + "options": [ + "A. Logistic regression", + "B. Random Cut Forest (RCF)", + "C. Principal component analysis (PCA)", + "D. Linear regression" + ], + "correct": "D. Linear regression", + "explanation": "Linear regression is a machine learning approach th at can be used to solve this problem. Linear regression is a supervised learning technique that can model the relationship between one or more input variables (features) and an output variable ( target). In this case, the input variables could be the historical sales data of the part, such as the quarter, the demand, the price, the inventory, etc. The output variable could be the number of units to be produced for the part. Linear regression can learn the coefficients (weights) of the input varia bles that best fit the output variable, and then us e them to make predictions for new data. Linear regre ssion is suitable for problems that involve continuous and numeric output variables, such as pr edicting house prices, stock prices, or sales volumes. Reference: AWS Machine Learning Specialty Exam Guide Linear Regression", + "references": "" + }, + { + "question": "A manufacturing company has structured and unstruct ured data stored in an Amazon S3 bucket A Machine Learning Specialist wants to use SQL to run queries on this dat", + "options": [ + "A. Which solution requires the LEAST effort to be ab le to query this data?", + "B. Use AWS Data Pipeline to transform the data and A mazon RDS to run queries.", + "C. Use AWS Glue to catalogue the data and Amazon Ath ena to run queries", + "D. Use AWS Batch to run ETL on the data and Amazon A urora to run the quenes", + "A. Bundle the NVIDIA drivers with the Docker image", + "B. Build the Docker container to be NVIDIA-Docker co mpatible", + "C. Organize the Docker container's file structure to execute on GPU instances.", + "D. Set the GPU flag in the Amazon SageMaker Create T rainingJob request body" + ], + "correct": "B. Build the Docker container to be NVIDIA-Docker co mpatible", + "explanation": "To leverage the NVIDIA GPUs on Amazon EC2 P3 instan ces, the Machine Learning Specialist needs to build the Docker container to be NVIDIA-Docker comp atible. NVIDIA-Docker is a tool that enables GPU-accelerated containers to run on Docker. It aut omatically configures the container to access the NVIDIA drivers and libraries on the host system. Th e Specialist does not need to bundle the NVIDIA drivers with the Docker image, as they are already installed on the EC2 P3 instances. The Specialist does not need to organize the Docker containers fil e structure to execute on GPU instances, as this is not relevant for GPU compatibility. The Specialist does not need to set the GPU flag in the Amazon SageMaker Create TrainingJob request body, as this is only required for using Elastic Inference accelerators, not EC2 P3 instances. Reference: NVID IA-Docker, Using GPU-Accelerated Containers, Using Elastic Inference in Amazon SageM aker", + "references": "" + }, + { + "question": "A large JSON dataset for a project has been uploade d to a private Amazon S3 bucket The Machine Learning Specialist wants to securely access and ex plore the data from an Amazon SageMaker notebook instance A new VPC was created and assigne d to the Specialist How can the privacy and integrity of the data store d in Amazon S3 be maintained while granting access to the Specialist for analysis?", + "options": [ + "A. Launch the SageMaker notebook instance within the VPC with SageMaker-provided internet", + "B. Launch the SageMaker notebook instance within the VPC and create an S3 VPC endpoint for the", + "C. Launch the SageMaker notebook instance within the VPC and create an S3 VPC endpoint for the", + "D. Launch the SageMaker notebook instance within the VPC with SageMaker-provided internet" + ], + "correct": "C. Launch the SageMaker notebook instance within the VPC and create an S3 VPC endpoint for the", + "explanation": "The best way to maintain the privacy and integrity of the data stored in Amazon S3 is to use a combination of VPC endpoints and S3 bucket policies . A VPC endpoint allows the SageMaker notebook instance to access the S3 bucket without g oing through the public internet. A bucket policy allows the S3 bucket owner to specify which VPCs or VPC endpoints can access the bucket. This way, the data is protected from unauthorized access and tampering. The other options are either insecure (A and D) or inefficient (B). Reference: Using Amaz on S3 VPC Endpoints, Using Bucket Policies and User Policies", + "references": "" + }, + { + "question": "Given the following confusion matrix for a movie cl assification model, what is the true class frequency for Romance and the predicted class frequ ency for Adventure?", + "options": [ + "A. The true class frequency for Romance is 77.56% an d the predicted class frequency for Adventure is", + "B. The true class frequency for Romance is 57.92% an d the predicted class frequency for Adventure is", + "C. The true class frequency for Romance is 0 78 and the predicted class frequency for Adventure is (0", + "D. The true class frequency for Romance is 77.56% * 0.78 and the predicted class frequency for" + ], + "correct": "B. The true class frequency for Romance is 57.92% an d the predicted class frequency for Adventure is", + "explanation": "The true class frequency for Romance is the percent age of movies that are actually Romance out of all the movies. This can be calculated by dividing the sum of the true values for Romance by the total number of movies. The predicted class frequency for Adventure is the percentage of movies that are predicted to be Adventure out of all the movies. Th is can be calculated by dividing the sum of the predicted values for Adventure by the total number of movies. Based on the confusion matrix, the true class frequency for Romance is 57.92% and the predicted class frequency for Adventure is 13.12%. Reference: Confusion Matrix, Classification Metrics", + "references": "" + }, + { + "question": "A Machine Learning Specialist is building a supervi sed model that will evaluate customers' satisfaction with their mobile phone service based on recent usage The model's output should infer whether or not a customer is likely to switch to a competitor in the next 30 days Which of the following modeling techniques should t he Specialist use1?", + "options": [ + "A. Time-series prediction", + "B. Anomaly detection", + "C. Binary classification", + "D. Regression" + ], + "correct": "C. Binary classification", + "explanation": "The modeling technique that the Machine Learning Sp ecialist should use is binary classification. Binary classification is a type of supervised learn ing that predicts whether an input belongs to one o f two possible classes. In this case, the input is th e customers recent usage data and the output is whether or not the customer is likely to switch to a competitor in the next 30 days. This is a binary outcome, either yes or no, so binary classification is suitable for this problem. The other options ar e not appropriate for this problem. Time-series predi ction is a type of supervised learning that forecasts future values based on past and present d ata. Anomaly detection is a type of unsupervised learning that identifies outliers or abnormal patte rns in the data. Regression is a type of supervised learning that estimates a continuous numerical valu e based on the input features. Reference: Binary Classification, Time Series Prediction, Anomaly Det ection, Regression", + "references": "" + }, + { + "question": "A web-based company wants to improve its conversion rate on its landing page Using a large historical dataset of customer visits, the company has repeatedly trained a multi-class deep learning network algorithm on Amazon SageMaker However there is an overfitting problem training data shows 90% accuracy in predictions, while test data shows 70% accuracy only The company needs to boost the generalization of it s model before deploying it into production to maximize conversions of visits to purchases Which action is recommended to provide the HIGHEST accuracy model for the company's test and validation data?", + "options": [ + "A. Increase the randomization of training data in th e mini-batches used in training.", + "B. Allocate a higher proportion of the overall data to the training dataset", + "C. Apply L1 or L2 regularization and dropouts to the training.", + "D. Reduce the number of layers and units (or neurons ) from the deep learning network." + ], + "correct": "C. Apply L1 or L2 regularization and dropouts to the training.", + "explanation": "Regularization and dropouts are techniques that can help reduce overfitting in deep learning models. Overfitting occurs when the model learns to o much from the training data and fails to generalize well to new data. Regularization adds a penalty term to the loss function that penalizes the model for having large or complex weights. This prevents the model from memorizing the noise or irrelevant features in the training data. L1 and L2 are two types of regularization that differ in how they calculate the penalty term. L1 regularization uses the absolute value of the weights, while L2 regularization uses the square of the weights. Drop outs are another technique that randomly drops out some units or neurons from the network during t raining. This creates a thinner network that is less prone to overfitting. Dropouts also act as a f orm of ensemble learning, where multiple sub- models are combined to produce a better prediction. By applying regularization and dropouts to the training, the web-based company can improve the gen eralization and accuracy of its deep learning model on the test and validation data. Reference: Regularization: A video that explains the concept a nd benefits of regularization in deep learning. Dropout: A video that demonstrates how dropout work s and why it helps reduce overfitting.", + "references": "" + }, + { + "question": "A Machine Learning Specialist was given a dataset c onsisting of unlabeled data The Specialist must create a model that can help the team classify the data into different buckets What model should be used to complete this work?", + "options": [ + "A. K-means clustering", + "B. Random Cut Forest (RCF)", + "C. XGBoost", + "D. BlazingText" + ], + "correct": "A. K-means clustering", + "explanation": "K-means clustering is a machine learning technique that can be used to classify unlabeled data into different groups based on their similarity. It is a n unsupervised learning method, which means it does not require any prior knowledge or labels for the data. K-means clustering works by randomly assigning data points to a number of clusters, then iteratively updating the cluster centers and reassigning the data points until the clusters are stable. The result is a partition of the data into distinct and homogeneous groups. K-means clustering can be useful for exploratory data analysis, data compression, anomaly detection, and feature ex traction. Reference: K-Means Clustering: A tutorial on how to use K-mean s clustering with Amazon SageMaker. Unsupervised Learning: A video that explains the co ncept and applications of unsupervised learning.", + "references": "" + }, + { + "question": "A retail company intends to use machine learning to categorize new products A labeled dataset of current products was provided to the Data Science t eam The dataset includes 1 200 products The labeled dataset has 15 features for each product su ch as title dimensions, weight, and price Each product is labeled as belonging to one of six categ ories such as books, games, electronics, and movies. Which model should be used for categorizing new pro ducts using the provided dataset for training?", + "options": [ + "A. An XGBoost model where the objective parameter is set to multi: softmax", + "B. A deep convolutional neural network (CNN) with a softmax activation function for the last layer", + "C. A regression forest where the number of trees is set equal to the number of product categories", + "D. A DeepAR forecasting model based on a recurrent n eural network (RNN)" + ], + "correct": "A. An XGBoost model where the objective parameter is set to multi: softmax", + "explanation": "XGBoost is a machine learning framework that can be used for classification, regression, ranking, and other tasks. It is based on the gradient boosting a lgorithm, which builds an ensemble of weak learners (usually decision trees) to produce a stro ng learner. XGBoost has several advantages over other algorithms, such as scalability, parallelizat ion, regularization, and sparsity handling. For categorizing new products using the provided datase t, an XGBoost model would be a suitable choice, because it can handle multiple features and multipl e classes efficiently and accurately. To train an XGBoost model for multi-class classification, the o bjective parameter should be set to multi: softmax, which means that the model will output a probabilit y distribution over the classes and predict the class with the highest probability. Alternatively, the objective parameter can be set to multi: softprob, which means that the model will output th e raw probability of each class instead of the predicted class label. This can be useful for evalu ating the model performance or for post-processing the predictions. Reference: XGBoost: A tutorial on how to use XGBoost with Amaz on SageMaker. XGBoost Parameters: A reference guide for the param eters of XGBoost.", + "references": "" + }, + { + "question": "A Machine Learning Specialist is building a model t o predict future employment rates based on a wide range of economic factors While exploring the data, the Specialist notices that the magnitude of the input features vary greatly The Specialist does not want variables with a larger magnitude to dominate the model What should the Specialist do to prepare the data f or model training'?", + "options": [ + "A. Apply quantile binning to group the data into cat egorical bins to keep any relationships in the data", + "B. Apply the Cartesian product transformation to cre ate new combinations of fields that are", + "C. Apply normalization to ensure each field will hav e a mean of 0 and a variance of 1 to remove any", + "D. Apply the orthogonal sparse Diagram (OSB) transfo rmation to apply a fixed-size sliding window to" + ], + "correct": "C. Apply normalization to ensure each field will hav e a mean of 0 and a variance of 1 to remove any", + "explanation": "Normalization is a data preprocessing technique tha t can be used to scale the input features to a common range, such as [-1, 1] or [0, 1]. Normalizat ion can help reduce the effect of outliers, improve the convergence of gradient-based algorithms, and p revent variables with a larger magnitude from dominating the model. One common method of normaliz ation is standardization, which transforms each feature to have a mean of 0 and a variance of 1. This can be done by subtracting the mean and dividing by the standard deviation of each feature. Standardization can be useful for models that assume the input features are normally distributed, such as linear regression, logistic regression, an d support vector machines. Reference: Data normalization and standardization: A video tha t explains the concept and benefits of data normalization and standardization. Standardize or Normalize?: A blog post that compare s different methods of scaling the input features.", + "references": "" + }, + { + "question": "A Machine Learning Specialist prepared the followin g graph displaying the results of k-means for k = [1:10] Considering the graph, what is a reasonable selecti on for the optimal choice of k?", + "options": [ + "A. 1", + "B. 4", + "C. 7", + "D. 10" + ], + "correct": "B. 4", + "explanation": "The elbow method is a technique that we use to dete rmine the number of centroids (k) to use in a kmean s clustering algorithm. In this method, we plot the w ithin-cluster sum of squares (WCSS) against the number of clusters (k) and look for the point where the curve bends sharply. This point is called the elbow point and it indicates that adding more clusters does not improve the model significantly. The graph in the question shows that the elbow point is at k = 4, which means that 4 is a reasonable choice for the optimal number of cluster s. Reference: Elbow Method for optimal value of k in KMeans: A tu torial on how to use the elbow method with Amazon SageMaker. K-Means Clustering: A video that explains the conce pt and benefits of k-means clustering.", + "references": "" + }, + { + "question": "A company is using Amazon Polly to translate plaint ext documents to speech for automated company announcements However company acronyms are being mispronounced in the current documents How should a Machine Learning Specialist address this issue for future documents?", + "options": [ + "A. Convert current documents to SSML with pronunciat ion tags", + "B. Create an appropriate pronunciation lexicon.", + "C. Output speech marks to guide in pronunciation", + "D. Use Amazon Lex to preprocess the text files for p ronunciation" + ], + "correct": "B. Create an appropriate pronunciation lexicon.", + "explanation": "A pronunciation lexicon is a file that defines how words or phrases should be pronounced by Amazon Polly. A lexicon can help customize the speech outp ut for words that are uncommon, foreign, or have multiple pronunciations. A lexicon must conform to the Pronunciation Lexicon Specification (PLS) standard and can be stored in an AWS region using t he Amazon Polly API. To use a lexicon for synthesizing speech, the lexicon name must be speci fied in the SSML tag. For example, the following lexicon defines how to pronounce the acro nym W3C: W3C World Wide Web Consortium To use this lexicon, the text input must include th e following SSML tag: The W3C is an international community that develops open st andards to ensure the long-term growth of the Web. ", + "references": "Customize pronunciation using lexicons in Amazon Po lly: A blog post that explains how to use lexicons for creating custom pronunciations. Managing Lexicons: A documentation page that descri bes how to store and retrieve lexicons using the Amazon Polly API." + }, + { + "question": "A Machine Learning Specialist is using Apache Spark for pre-processing training data As part of the Spark pipeline, the Specialist wants to use Amazon SageMaker for training a model and hosting it Which of the following would the Specialist do to i ntegrate the Spark application with SageMaker? (Select THREE)", + "options": [ + "A. Download the AWS SDK for the Spark environment", + "B. Install the SageMaker Spark library in the Spark environment.", + "C. Use the appropriate estimator from the SageMaker Spark Library to train a model.", + "D. Compress the training data into a ZIP file and up load it to a pre-defined Amazon S3 bucket." + ], + "correct": "", + "explanation": "The SageMaker Spark library is a library that enabl es Apache Spark applications to integrate with Amazon SageMaker for training and hosting machine l earning models. The library provides several features, such as: Estimators: Classes that allow Spark users to train Amazon SageMaker models and host them on Amazon SageMaker endpoints using the Spark MLlib Pi pelines API. The library supports various builtin algorithms, such as linear learner, XGBoost, K-mean s, etc., as well as custom algorithms using Docker containers. Model classes: Classes that wrap Amazon SageMaker m odels in a Spark MLlib Model abstraction. This allows Spark users to use Amazon SageMaker end points for inference within Spark applications. Data sources: Classes that allow Spark users to rea d data from Amazon S3 using the Spark Data Sources API. The library supports various data form ats, such as CSV, LibSVM, RecordIO, etc. To integrate the Spark application with SageMaker, the Machine Learning Specialist should do the following: Install the SageMaker Spark library in the Spark en vironment. This can be done by using Maven, pip, or downloading the JAR file from GitHub. Use the appropriate estimator from the SageMaker Sp ark Library to train a model. For example, to train a linear learner model, the Specialist can us e the following code: Use the sageMakerModel. transform method to get inf erences from the model hosted in SageMaker. For example, to get predictions for a test DataFram e, the Specialist can use the following code:", + "references": "[SageMaker Spark]: A documentation page that introd uces the SageMaker Spark library and its features. [SageMaker Spark GitHub Repository]: A GitHub repos itory that contains the source code, examples, and installation instructions for the SageMaker Spa rk library." + }, + { + "question": "A Machine Learning Specialist is working with a lar ge cybersecurily company that manages security events in real time for companies around the world The cybersecurity company wants to design a solution that will allow it to use machine learning to score malicious events as anomalies on the data as it is being ingested The company also wants be a ble to save the results in its data lake for later processing and analysis What is the MOST efficient way to accomplish these tasks'?", + "options": [ + "A. Ingest the data using Amazon Kinesis Data Firehos e, and use Amazon Kinesis Data Analytics", + "B. Ingest the data into Apache Spark Streaming using Amazon EMR. and use Spark MLlib with kmeans", + "C. Ingest the data and store it in Amazon S3 Use AWS Batch along with the AWS Deep Learning AMIs", + "D. Ingest the data and store it in Amazon S3. Have a n AWS Glue job that is triggered on demand" + ], + "correct": "A. Ingest the data using Amazon Kinesis Data Firehos e, and use Amazon Kinesis Data Analytics", + "explanation": "Amazon Kinesis Data Firehose is a fully managed ser vice that can capture, transform, and load streaming data into AWS data stores, such as Amazon S3, Amazon Redshift, Amazon Elasticsearch Service, and Splunk. It can also invoke AWS Lambda functions to perform custom transformations on the data. Amazon Kinesis Data Analytics is a servic e that can analyze streaming data in real time using SQL or Apache Flink applications. It can also use machine learning algorithms, such as Random Cut Forest (RCF), to perform anomaly detection on s treaming data. RCF is an unsupervised learning algorithm that assigns an anomaly score to each dat a point based on how different it is from the rest of the data. By using Kinesis Data Firehose and Kin esis Data Analytics, the cybersecurity company can ingest the data in real time, score the malicious e vents as anomalies, and stream the results to Amazon S3, which can serve as a data lake for later processing and analysis. This is the most efficien t way to accomplish these tasks, as it does not requi re any additional infrastructure, coding, or training.", + "references": "Amazon Kinesis Data Firehose - Amazon Web Services Amazon Kinesis Data Analytics - Amazon Web Services Anomaly Detection with Amazon Kinesis Data Analytic s - Amazon Web Services [AWS Certified Machine Learning - Specialty Sample Questions]" + }, + { + "question": "A Machine Learning Specialist works for a credit ca rd processing company and needs to predict which transactions may be fraudulent in near-real time. S pecifically, the Specialist must train a model that returns the probability that a given transaction ma y be fraudulent How should the Specialist frame this business probl em'?", + "options": [ + "A. Streaming classification", + "B. Binary classification", + "C. Multi-category classification", + "D. Regression classification" + ], + "correct": "B. Binary classification", + "explanation": "Binary classification is a type of supervised learn ing problem where the goal is to predict a categorical label that has only two possible values , such as Yes or No, True or False, Positive or Negative. In this case, the label is whether a tran saction is fraudulent or not, which is a binary outcome. Binary classification can be used to estim ate the probability of an observation belonging to a certain class, such as the probability of a trans action being fraudulent. This can help the business to make decisions based on the risk level of each tran saction. Reference: Binary Classification - Amazon Machine Learning AWS Certified Machine Learning - Specialty Sample Q uestions", + "references": "" + }, + { + "question": "Amazon Connect has recently been tolled out across a company as a contact call center The solution has been configured to store voice call recordings on Amazon S3 The content of the voice calls are being analyzed f or the incidents being discussed by the call operators Amazon Transcribe is being used to conver t the audio to text, and the output is stored on Amazon S3 Which approach will provide the information require d for further analysis?", + "options": [ + "A. Use Amazon Comprehend with the transcribed files to build the key topics", + "B. Use Amazon Translate with the transcribed files t o train and build a model for the key topics", + "C. Use the AWS Deep Learning AMI with Gluon Semantic Segmentation on the transcribed files to", + "D. Use the Amazon SageMaker k-Nearest-Neighbors (kNN ) algorithm on the transcribed files to" + ], + "correct": "A. Use Amazon Comprehend with the transcribed files to build the key topics", + "explanation": "Amazon Comprehend is a natural language processing (NLP) service that uses machine learning to find insights and relationships in text. It can ana lyze text documents and identify the key topics, entities, sentiments, languages, and more. In this case, Amazon Comprehend can be used with the transcribed files from Amazon Transcribe to extract the main topics that are being discussed by the call operators. This can help to understand the com mon issues and concerns of the customers, and provide insights for further analysis and improveme nt. Reference: Amazon Comprehend - Amazon Web Services AWS Certified Machine Learning - Specialty Sample Q uestions", + "references": "" + }, + { + "question": "A Machine Learning Specialist is building a predict ion model for a large number of features using linear models, such as linear regression and logist ic regression During exploratory data analysis the Specialist observes that many features are highly c orrelated with each other This may make the model unstable What should be done to reduce the impact of having such a large number of features?", + "options": [ + "A. Perform one-hot encoding on highly correlated fea tures", + "B. Use matrix multiplication on highly correlated fe atures.", + "C. Create a new feature space using principal compon ent analysis (PCA)", + "D. Apply the Pearson correlation coefficient" + ], + "correct": "C. Create a new feature space using principal compon ent analysis (PCA)", + "explanation": "Principal component analysis (PCA) is an unsupervis ed machine learning algorithm that attempts to reduce the dimensionality (number of features) with in a dataset while still retaining as much information as possible. This is done by finding a new set of features called components, which are composites of the original features that are uncorr elated with one another. They are also constrained so that the first component accounts for the larges t possible variability in the data, the second component the second most variability, and so on. B y using PCA, the impact of having a large number of features that are highly correlated with each other can be reduced, as the new feature space will have fewer dimensions and less redundanc y. This can make the linear models more stable and less prone to overfitting. Reference: Principal Component Analysis (PCA) Algorithm - Amaz on SageMaker Perform a large-scale principal component analysis faster using Amazon SageMaker | AWS Machine Learning Blog Machine Learning- Prinicipal Component Analysis | i 2tutorials", + "references": "" + }, + { + "question": "A Machine Learning Specialist wants to determine th e appropriate SageMaker Variant Invocations Per Instance setting for an endpoint automatic scal ing configuration. The Specialist has performed a load test on a single instance and determined that peak requests per second (RPS) without service degradation is about 20 RPS As this is the first de ployment, the Specialist intends to set the invocation safety factor to 0 5 Based on the stated parameters and given that the i nvocations per instance setting is measured on a per-minute basis, what should the Specialist set as the sageMaker variant invocations Per instance setting?", + "options": [ + "A. 10", + "B. 30", + "C. 600", + "D. 2,400" + ], + "correct": "C. 600", + "explanation": "The SageMaker Variant Invocations Per Instance sett ing is the target value for the average number of invocations per instance per minute for the model v ariant. It is used by the automatic scaling policy to add or remove instances to keep the metric close to the specified value. To determine this value, the following equation can be used in combination w ith load testing: SageMakerVariantInvocationsPerInstance = (MAX_RPS * SAFETY_FACTOR) * 60 Where MAX_RPS is the maximum requests per second th at the model variant can handle without service degradation, SAFETY_FACTOR is a factor that ensures that the clients do not exceed the maximum RPS, and 60 is the conversion factor from s econds to minutes. In this case, the given parameters are: MAX_RPS = 20 SAFETY_FACTOR = 0.5 Plugging these values into the equation, we get: SageMakerVariantInvocationsPerInstance = (20 * 0.5) * 60 SageMakerVariantInvocationsPerInstance = 600 Therefore, the Specialist should set the SageMaker Variant Invocations Per Instance setting to 600.", + "references": "Load testing your auto scaling configuration - Amaz on SageMaker Configure model auto scaling with the console - Ama zon SageMaker" + }, + { + "question": "A Machine Learning Specialist deployed a model that provides product recommendations on a company's website Initially, the model was performi ng very well and resulted in customers buying more products on average However within the past fe w months the Specialist has noticed that the effect of product recommendations has diminished an d customers are starting to return to their original habits of spending less The Specialist is unsure of what happened, as the model has not changed from its initial deployment over a year agoWhich method should the Specialist try to improve m odel performance?", + "options": [ + "A. The model needs to be completely re-engineered be cause it is unable to handle product inventory", + "B. The model's hyperparameters should be periodicall y updated to prevent drift", + "C. The model should be periodically retrained from s cratch using the original data while adding a", + "D. The model should be periodically retrained using the original training data plus new data as" + ], + "correct": "D. The model should be periodically retrained using the original training data plus new data as", + "explanation": "The problem that the Machine Learning Specialist is facing is likely due to concept drift, which is a phenomenon where the statistical properties of the target variable change over time, making the model less accurate and relevant. Concept drift can occur due to various reasons, such as changes in customer preferences, market trends, product invent ory, seasonality, etc. In this case, the product recommendations model may have become outdated as t he product inventory changed over time, making the recommendations less appealing to the cu stomers. To address this issue, the model should be periodically retrained using the original training data plus new data as product inventory changes. This way, the model can learn from the lat est data and adapt to the changing customer behavior and preferences. Retraining the model from scratch using the original data while adding a regularization term may not be sufficient, as it do es not account for the new data. Updating the models hyperparameters may not help either, as it d oes not address the underlying data distribution change. Re-engineering the model completely may not be necessary, as the model may still be valid and useful with periodic retraining.", + "references": "Concept Drift - Amazon SageMaker Detecting and Handling Concept Drift - Amazon SageM aker Machine Learning Concepts - Amazon Machine Learning" + }, + { + "question": "A manufacturer of car engines collects data from ca rs as they are being driven The data collected includes timestamp, engine temperature, rotations p er minute (RPM), and other sensor readings The company wants to predict when an engine is going to have a problem so it can notify drivers in advance to get engine maintenance The engine data i s loaded into a data lake for training Which is the MOST suitable predictive model that ca n be deployed into production'?", + "options": [ + "A. Add labels over time to indicate which engine fau lts occur at what time in the future to turn this", + "B. This data requires an unsupervised learning algor ithm Use Amazon SageMaker k-means to cluster", + "C. Add labels over time to indicate which engine fau lts occur at what time in the future to turn this", + "D. This data is already formulated as a time series Use Amazon SageMaker seq2seq to model the" + ], + "correct": "A. Add labels over time to indicate which engine fau lts occur at what time in the future to turn this", + "explanation": "A recurrent neural network (RNN) is a type of neura l network that can process sequential data, such as time series, by maintaining a hidden state that captures the temporal dependencies between the inputs. RNNs are well suited for predicting future events based on past observations, such as forecasting engine failures based on sensor reading s. To train an RNN model, the data needs to be labeled with the target variable, which in this cas e is the type and time of the engine fault. This makes the problem a supervised learning problem, wh ere the goal is to learn a mapping from the input sequence (sensor readings) to the output sequ ence (engine faults). By using an RNN model, the manufacturer can leverage the temporal information in the data and detect patterns that indicate when an engine might need maintenance for a certain fault.", + "references": "Recurrent Neural Networks - Amazon SageMaker Use Amazon SageMaker Built-in Algorithms or Pre-tra ined Models Recurrent Neural Network Definition | DeepAI What are Recurrent Neural Networks? An Ultimate Gui de for Newbies! Lee and Carter go Machine Learning: Recurrent Neura l Networks - SSRN" + }, + { + "question": "A Data Scientist is working on an application that performs sentiment analysis. The validation accuracy is poor and the Data Scientist thinks that the cause may be a rich vocabulary and a low average frequency of words in the dataset Which tool should be used to improve the validation accuracy?", + "options": [ + "A. Amazon Comprehend syntax analysts and entity dete ction", + "B. Amazon SageMaker BlazingText allow mode", + "C. Natural Language Toolkit (NLTK) stemming and stop word removal", + "D. Scikit-learn term frequency-inverse document freq uency (TF-IDF) vectorizers" + ], + "correct": "D. Scikit-learn term frequency-inverse document freq uency (TF-IDF) vectorizers", + "explanation": "Term frequency-inverse document frequency (TF-IDF) is a technique that assigns a weight to each word in a document based on how important it is to the meaning of the document. The term frequency (TF) measures how often a word appears in a document, while the inverse document frequency (IDF) measures how rare a word is across a collection of documents. The TF-IDF weight is the product of the TF and IDF values, and it is hig h for words that are frequent in a specific documen t but rare in the overall corpus. TF-IDF can help imp rove the validation accuracy of a sentiment analysis model by reducing the impact of common wor ds that have little or no sentiment value, such as oethe , oea , oeand , etc. Scikit-learn is a pop ular Python library for machine learning that provi des a TF-IDF vectorizer class that can transform a collec tion of text documents into a matrix of TF-IDF features. By using this tool, the Data Scientist ca n create a more informative and discriminative feature representation for the sentiment analysis t ask.", + "references": "TfidfVectorizer - scikit-learn Text feature extraction - scikit-learn TF-IDF for Beginners | by Jana Schmidt | Towards Da ta Science Sentiment Analysis: Concept, Analysis and Applicati ons | by Susan Li | Towards Data Science" + }, + { + "question": "A Machine Learning Specialist is developing recomme ndation engine for a photography blog Given a picture, the recommendation engine should show a pi cture that captures similar objects The Specialist would like to create a numerical represe ntation feature to perform nearest-neighbor searches What actions would allow the Specialist to get rele vant numerical representations?", + "options": [ + "A. Reduce image resolution and use reduced resolution pixel values as features B. Use Amazon Mechanical Turk to label image content and create a one-hot representation", + "C. Run images through a neural network pie-trained o n ImageNet, and collect the feature vectors", + "D. Average colors by channel to obtain three-dimensi onal representations of images." + ], + "correct": "C. Run images through a neural network pie-trained o n ImageNet, and collect the feature vectors", + "explanation": "A neural network pre-trained on ImageNet is a deep learning model that has been trained on a large dataset of images containing 1000 classes of object s. The model can learn to extract high-level features from the images that capture the semantic and visual information of the objects. The penultimate layer of the model is the layer before the final output layer, and it contains a feature vector that represents the input image in a lower-d imensional space. By running images through a pre-trained neural network and collecting the featu re vectors from the penultimate layer, the Specialist can obtain relevant numerical representa tions that can be used for nearest-neighbor searches. The feature vectors can capture the simil arity between images based on the presence and appearance of similar objects, and they can be comp ared using distance metrics such as Euclidean distance or cosine similarity. This approach can en able the recommendation engine to show a picture that captures similar objects to a given picture.", + "references": "ImageNet - Wikipedia How to use a pre-trained neural network to extract features from images | by Rishabh Anand | Analytics Vidhya | Medium Image Similarity using Deep Ranking | by Aditya Oke | Towards Data Science" + }, + { + "question": "A gaming company has launched an online game where people can start playing for free but they need to pay if they choose to use certain features The company needs to build an automated system to predict whether or not a new user will become a paid user within 1 year The company has gathered a labeled dataset from 1 million users The training dataset consists of 1.000 positive sam ples (from users who ended up paying within 1 year) and 999.000 negative samples (from users who did not use any paid features) Each data sample consists of 200 features including user age, device , location, and play patterns Using this dataset for training, the Data Science t eam trained a random forest model that converged with over 99% accuracy on the training set However, the prediction results on a test dataset were not satisfactory. Which of the following approaches should the Data S cience team take to mitigate this issue? (Select TWO.)", + "options": [ + "A. Add more deep trees to the random forest to enabl e the model to learn more features.", + "B. indicate a copy of the samples in the test databa se in the training dataset", + "C. Generate more positive samples by duplicating the positive samples and adding a small amount of", + "D. Change the cost function so that false negatives have a higher impact on the cost value than false" + ], + "correct": "", + "explanation": "The Data Science team is facing a problem of imbala nced data, where the positive class (paid users) is much less frequent than the negative class (non- paid users). This can cause the random forest model to be biased towards the majority class and h ave poor performance on the minority class. To mitigate this issue, the Data Science team can try the following approaches: C) Generate more positive samples by duplicating th e positive samples and adding a small amount of noise to the duplicated data. This is a technique c alled data augmentation, which can help increase the size and diversity of the training data for the minority class. This can help the random forest model learn more features and patterns from the pos itive class and reduce the imbalance ratio. D) Change the cost function so that false negatives have a higher impact on the cost value than false positives. This is a technique called cost-sensitiv e learning, which can assign different weights or costs to different classes or errors. By assigning a higher cost to false negatives (predicting non-pa id when the user is actually paid), the random forest model can be more sensitive to the minority class and try to minimize the misclassification of the po sitive class.", + "references": "Bagging and Random Forest for Imbalanced Classifica tion Surviving in a Random Forest with Imbalanced Datase ts machine learning - random forest for imbalanced dat a? - Cross Validated Biased Random Forest For Dealing With the Class Imb alance Problem" + }, + { + "question": "While reviewing the histogram for residuals on regr ession evaluation data a Machine Learning Specialist notices that the residuals do not form a zero-centered bell shape as shown What does this mean?", + "options": [ + "A. The model might have prediction errors over a ran ge of target values.", + "B. The dataset cannot be accurately represented usin g the regression model", + "C. There are too many variables in the model", + "D. The model is predicting its target values perfect ly." + ], + "correct": "A. The model might have prediction errors over a ran ge of target values.", + "explanation": "Residuals are the differences between the actual an d predicted values of the target variable in a regression model. A histogram of residuals is a gra phical tool that can help evaluate the performance and assumptions of the model. Ideally, the histogra m of residuals should have a zero-centered bell shape, which indicates that the residuals are norma lly distributed with a mean of zero and a constant variance. This means that the model has captured th e true relationship between the input and output variables, and that the errors are random an d unbiased. However, if the histogram of residuals does not have a zero-centered bell shape, as shown in the image, this means that the model might have prediction errors over a range of target values. This is because the residuals do not form a symmetrical and homogeneous distribution aro und zero, which implies that the model has some systematic bias or heteroscedasticity. This ca n affect the accuracy and validity of the model, and indicate that the model needs to be improved or modified.", + "references": "Residual Analysis in Regression - Statistics By Jim How to Check Residual Plots for Regression Analysis - dummies Histogram of Residuals - Statistics How To" + }, + { + "question": "During mini-batch training of a neural network for a classification problem, a Data Scientist notices that training accuracy oscillates What is the MOST likely cause of this issue?", + "options": [ + "A. The class distribution in the dataset is imbalanc ed", + "B. Dataset shuffling is disabled", + "C. The batch size is too big", + "D. The learning rate is very high" + ], + "correct": "D. The learning rate is very high", + "explanation": "Mini-batch gradient descent is a variant of gradien t descent that updates the model parameters using a subset of the training data (called a mini- batch) at each iteration. The learning rate is a hyperparameter that controls how much the model par ameters change in response to the gradient. If the learning rate is very high, the model parame ters may overshoot the optimal values and oscillate around the minimum of the cost function. This can cause the training accuracy to fluctuate and prevent the model from converging to a stable s olution. To avoid this issue, the learning rate should be chosen carefully, such as by using a lear ning rate decay schedule or an adaptive learning rate algorithm1. Alternatively, the batch size can be increased to reduce the variance of the gradient estimates2. However, the batch size should not be t oo big, as this can slow down the training process and reduce the generalization ability of the model3 . Dataset shuffling and class distribution are not likely to cause oscillations in training accuracy, as they do not affect the gradient updates directly. Dataset shuffling can help avoid getting stuck in local minima and improve the convergence speed of mini-batch gradient descent4. Class distri bution can affect the performance and fairness of the model, especially if the dataset is imbalanced, but it does not necessarily cause fluctuations in training accuracy.", + "references": "" + }, + { + "question": "A Machine Learning Specialist observes several perf ormance problems with the training portion of a machine learning solution on Amazon SageMaker The s olution uses a large training dataset 2 TB in size and is using the SageMaker k-means algorithm T he observed issues include the unacceptable length of time it takes before the training job lau nches and poor I/O throughput while training the model What should the Specialist do to address the perfor mance issues with the current solution? A. Use the SageMaker batch transform feature", + "options": [ + "B. Compress the training data into Apache Parquet fo rmat.", + "C. Ensure that the input mode for the training job i s set to Pipe.", + "D. Copy the training dataset to an Amazon EFS volume mounted on the SageMaker instance." + ], + "correct": "C. Ensure that the input mode for the training job i s set to Pipe.", + "explanation": "The input mode for the training job determines how the training data is transferred from Amazon S3 to the SageMaker instance. There are two input mode s: File and Pipe. File mode copies the entire training dataset from S3 to the local file system o f the instance before starting the training job. Th is can cause a long delay before the training job laun ches, especially if the dataset is large. Pipe mode streams the data from S3 to the instance as the tra ining job runs. This can reduce the startup time and improve the I/O throughput, as the data is read in smaller batches. Therefore, to address the performance issues with the current solution, the S pecialist should ensure that the input mode for the training job is set to Pipe. This can be done b y using the SageMaker Python SDK and setting the input_mode parameter to Pipe when creating the esti mator or the fit method12. Alternatively, this can be done by using the AWS CLI and setting the In putMode parameter to Pipe when creating the training job3.", + "references": "Access Training Data - Amazon SageMaker Choosing Data Input Mode Using the SageMaker Python SDK - Amazon SageMaker CreateTrainingJob - Amazon SageMaker Service" + }, + { + "question": "A Machine Learning Specialist is building a convolu tional neural network (CNN) that will classify 10 types of animals. The Specialist has built a series of layers in a neural network that will take an in put image of an animal, pass it through a series of con volutional and pooling layers, and then finally pas s it through a dense and fully connected layer with 1 0 nodes The Specialist would like to get an output from the neural network that is a probability distr ibution of how likely it is that the input image belongs to each of the 10 classes Which function will produce the desired output?", + "options": [ + "A. Dropout", + "B. Smooth L1 loss", + "C. Softmax", + "D. Rectified linear units (ReLU)" + ], + "correct": "C. Softmax", + "explanation": "The softmax function is a function that can transfo rm a vector of arbitrary real values into a vector of real values in the range (0,1) that sum to 1. This means that the softmax function can produce a valid probability distribution over multiple classes. The softmax function is often used as the activation function of the output layer in a neural network, e specially for multi-class classification problems. The softmax function can assign higher probabilitie s to the classes with higher scores, which allows the network to make predictions based on the most l ikely class. In this case, the Machine Learning Specialist wants to get an output from the neural n etwork that is a probability distribution of how likely it is that the input image belongs to each o f the 10 classes of animals. Therefore, the softmax function is the most suitable function to produce t he desired output.", + "references": "Softmax Activation Function for Deep Learning: A Co mplete Guide What is Softmax in Machine Learning? - reason.town machine learning - Why is the softmax function ofte n used as activation \u00a6 Multi-Class Neural Networks: Softmax | Machine Lear ning | Google for \u00a6" + }, + { + "question": "A Machine Learning Specialist is building a model t hat will perform time series forecasting using Amazon SageMaker The Specialist has finished traini ng the model and is now planning to perform load testing on the endpoint so they can configure Auto Scaling for the model variant Which approach will allow the Specialist to review the latency, memory utilization, and CPU utilization during the load test\"?", + "options": [ + "A. Review SageMaker logs that have been written to A mazon S3 by leveraging Amazon Athena and", + "B. Generate an Amazon CloudWatch dashboard to create a single view for the latency, memory", + "C. Build custom Amazon CloudWatch Logs and then leve rage Amazon ES and Kibana to query and", + "D. Send Amazon CloudWatch Logs that were generated b y Amazon SageMaker lo Amazon ES and use" + ], + "correct": "B. Generate an Amazon CloudWatch dashboard to create a single view for the latency, memory", + "explanation": "Amazon CloudWatch is a service that can monitor and collect various metrics and logs from AWS resources, such as Amazon SageMaker. Amazon CloudWa tch can also generate dashboards to create a single view for the metrics and logs that are of interest. By using Amazon CloudWatch, the Machine Learning Specialist can review the latency, memory utilization, and CPU utilization during the load test, as these are some of the metrics that are out putted by Amazon SageMaker. The Specialist can create a custom dashboard that displays these metri cs in different widgets, such as graphs, tables, or text. The dashboard can also be configured to refre sh automatically and show the latest data as the load test is running. This approach will allow the Specialist to monitor the performance and resource utilization of the model variant and adjust the Aut o Scaling configuration accordingly.", + "references": "[Monitoring Amazon SageMaker with Amazon CloudWatch - Amazon SageMaker] [Using Amazon CloudWatch Dashboards - Amazon CloudW atch] [Create a CloudWatch Dashboard - Amazon CloudWatch]" + }, + { + "question": "An Amazon SageMaker notebook instance is launched i nto Amazon VPC The SageMaker notebook references data contained in an Amazon S3 bucket in another account The bucket is encrypted using SSE-KMS The instance returns an access denied error when trying to access data in Amazon S3. Which of the following are required to access the b ucket and avoid the access denied error? (Select THREE)", + "options": [ + "A. An AWS KMS key policy that allows access to the c ustomer master key (CMK)", + "B. A SageMaker notebook security group that allows a ccess to Amazon S3", + "C. An 1AM role that allows access to the specific S3 bucket", + "D. A permissive S3 bucket policy" + ], + "correct": "", + "explanation": "To access an Amazon S3 bucket in another account th at is encrypted using SSE-KMS, the following are required: A) An AWS KMS key policy that allows access to the customer master key (CMK). The CMK is the encryption key that is used to encrypt and decrypt the data in the S3 bucket. The KMS key policy defines who can use and manage the CMK. To allow ac cess to the CMK from another account, the key policy must include a statement that grants the necessary permissions (such as kms:Decrypt) to the principal from the other account (such as the S ageMaker notebook IAM role). B) A SageMaker notebook security group that allows access to Amazon S3. A security group is a virtual firewall that controls the inbound and outb ound traffic for the SageMaker notebook instance. To allow the notebook instance to access the S3 buc ket, the security group must have a rule that allows outbound traffic to the S3 endpoint on port 443 (HTTPS). C) An IAM role that allows access to the specific S 3 bucket. An IAM role is an identity that can be assumed by the SageMaker notebook instance to acces s AWS resources. The IAM role must have a policy that grants the necessary permissions (such as s3:GetObject) to access the specific S3 bucket. The policy must also include a condition that allow s access to the CMK in the other account. The following are not required or correct: D) A permissive S3 bucket policy. A bucket policy i s a resource-based policy that defines who can access the S3 bucket and what actions they can perf orm. A permissive bucket policy is not required and not recommended, as it can expose the bucket to unauthorized access. A bucket policy should follow the principle of least privilege and grant t he minimum permissions necessary to the specific principals that need access. E) An S3 bucket owner that matches the notebook own er. The S3 bucket owner and the notebook owner do not need to match, as long as the bucket o wner grants cross-account access to the notebook owner through the KMS key policy and the b ucket policy (if applicable). F) A SegaMaker notebook subnet ACL that allow traff ic to Amazon S3. A subnet ACL is a network access control list that acts as an optional layer of security for the SageMaker notebook instances subnet. A subnet ACL is not required to access the S3 bucket, as the security group is sufficient to control the traffic. However, if a subnet ACL is us ed, it must not block the traffic to the S3 endpoin t.", + "references": "" + }, + { + "question": "A monitoring service generates 1 TB of scale metric s record data every minute A Research team performs queries on this data using Amazon Athena T he queries run slowly due to the large volume of data, and the team requires better performance How should the records be stored in Amazon S3 to im prove query performance?", + "options": [ + "A. CSV files", + "B. Parquet files", + "C. Compressed JSON", + "D. RecordIO" + ], + "correct": "B. Parquet files", + "explanation": "Parquet is a columnar storage format that can store data in a compressed and efficient way. Parquet files can improve query performance by reducing the amount of data that needs to be scanned, as only the relevant columns are read from the files. Parquet files can also support predicate pushdown, which means that the filtering conditions are appli ed at the storage level, further reducing the data that needs to be processed. Parquet files are compa tible with Amazon Athena, which can leverage the benefits of the columnar format and provide fas ter and cheaper queries. Therefore, the records should be stored in Parquet files in Amazon S3 to i mprove query performance.", + "references": "Columnar Storage Formats - Amazon Athena Parquet SerDe - Amazon Athena Optimizing Amazon Athena Queries - Amazon Athena Parquet - Apache Software Foundation" + }, + { + "question": "A Machine Learning Specialist needs to create a dat a repository to hold a large amount of time-based training data for a new model. In the source system , new files are added every hour Throughout a single 24-hour period, the volume of hourly updates will change significantly. The Specialist always wants to train on the last 24 hours of the data Which type of data repository is the MOST cost-effe ctive solution?", + "options": [ + "A. An Amazon EBS-backed Amazon EC2 instance with hou rly directories", + "B. An Amazon RDS database with hourly table partitio ns", + "C. An Amazon S3 data lake with hourly object prefixe s", + "D. An Amazon EMR cluster with hourly hive partitions on Amazon EBS volumes" + ], + "correct": "C. An Amazon S3 data lake with hourly object prefixe s", + "explanation": "An Amazon S3 data lake is a cost-effective solution for storing and analyzing large amounts of timebas ed training data for a new model. Amazon S3 is a highl y scalable, durable, and secure object storage service that can store any amount of data i n any format. Amazon S3 also offers low-cost storage classes, such as S3 Standard-IA and S3 One Zone-IA, that can reduce the storage costs for infrequently accessed data. By using hourly object prefixes, the Machine Learning Specialist can organize the data into logical partitions based on the time of ingestion. This can enable efficient da ta access and management, as well as support increment al updates and deletes. The Specialist can also use Amazon S3 lifecycle policies to automatically t ransition the data to lower-cost storage classes or delete the data after a certain period of time. Thi s way, the Specialist can always train on the last 24 hours of the data and optimize the storage costs.", + "references": "What is a data lake? - Amazon Web Services Amazon S3 Storage Classes - Amazon Simple Storage S ervice Managing your storage lifecycle - Amazon Simple Sto rage Service Best Practices Design Patterns: Optimizing Amazon S 3 Performance" + }, + { + "question": "A retail chain has been ingesting purchasing record s from its network of 20,000 stores to Amazon S3 using Amazon Kinesis Data Firehose To support train ing an improved machine learning model, training records will require new but simple transf ormations, and some attributes will be combined The model needs lo be retrained daily Given the large number of stores and the legacy dat a ingestion, which change will require the LEAST amount of development effort?", + "options": [ + "A. Require that the stores to switch to capturing th eir data locally on AWS Storage Gateway for", + "B. Deploy an Amazon EMR cluster running Apache Spark with the transformation logic, and have the", + "C. Spin up a fleet of Amazon EC2 instances with the transformation logic, have them transform the", + "D. Insert an Amazon Kinesis Data Analytics stream do wnstream of the Kinesis Data Firehouse stream" + ], + "correct": "D. Insert an Amazon Kinesis Data Analytics stream do wnstream of the Kinesis Data Firehouse stream", + "explanation": "Amazon Kinesis Data Analytics is a service that can analyze streaming data in real time using SQL or Apache Flink applications. It can also use machine learning algorithms, such as Random Cut Forest (RCF), to perform anomaly detection on streaming da ta. By inserting a Kinesis Data Analytics stream downstream of the Kinesis Data Firehose stream, the retail chain can transform the raw record attributes into simple transformed values using SQL queries. This can be done without changing the existing data ingestion process or deploying additi onal resources. The transformed records can then be outputted to another Kinesis Data Firehose strea m that delivers them to Amazon S3 for training the machine learning model. This approach will requ ire the least amount of development effort, as it leverages the existing Kinesis Data Firehose stream and the built-in SQL capabilities of Kinesis Data Analytics.", + "references": "Amazon Kinesis Data Analytics - Amazon Web Services Anomaly Detection with Amazon Kinesis Data Analytic s - Amazon Web Services Amazon Kinesis Data Firehose - Amazon Web Services Amazon S3 - Amazon Web Services" + }, + { + "question": "A city wants to monitor its air quality to address the consequences of air pollution A Machine Learning Specialist needs to forecast the air quali ty in parts per million of contaminates for the nex t 2 days in the city as this is a prototype, only daily data from the last year is available Which model is MOST likely to provide the best resu lts in Amazon SageMaker?", + "options": [ + "A. Use the Amazon SageMaker k-Nearest-Neighbors (kNN ) algorithm on the single time series", + "B. Use Amazon SageMaker Random Cut Forest (RCF) on t he single time series consisting of the full", + "C. Use the Amazon SageMaker Linear Learner algorithm on the single time series consisting of the", + "D. Use the Amazon SageMaker Linear Learner algorithm on the single time series consisting of the", + "A. Recall = 0.92 Precision = 0.84", + "B. Recall = 0.84 Precision = 0.8", + "C. Recall = 0.92 Precision = 0.8", + "D. Recall = 0.8 Precision = 0.92" + ], + "correct": "C. Recall = 0.92 Precision = 0.8", + "explanation": "Recall and precision are two metrics that can be us ed to evaluate the performance of a classification model. Recall is the ratio of true positives to the total number of actual positives, which measures how well the model can identify all the relevant ca ses. Precision is the ratio of true positives to th e total number of predicted positives, which measures how accurate the model is when it makes a positive prediction. Based on the confusion matrix in the image, we can calculate the recall and precision as follows: Recall = TP / (TP + FN) = 12 / (12 + 1) = 0.92 Precision = TP / (TP + FP) = 12 / (12 + 3) = 0.8 Where TP is the number of true positives, FN is the number of false negatives, and FP is the number of false positives. Therefore, the recall and preci sion of the model are 0.92 and 0.8, respectively.", + "references": "Amazon SageMaker k-Nearest-Neighbors (kNN) Algorith m - Amazon SageMaker Time Series Forecasting using k-Nearest Neighbors ( kNN) in Python | by \u00a6 Time Series Forecasting with k-Nearest Neighbors | by Nishant Malik \u00a6 QUESTION 38 For the given confusion matrix, what is the recall and precision of the model?" + }, + { + "question": "A Machine Learning Specialist is working with a med ia company to perform classification on popular articles from the company's website. The company is using random forests to classify how popular an article will be before it is published A sample of the data being used is below. Given the dataset, the Specialist wants to convert the Day-Of_Week column to binary values. What technique should be used to convert this colum n to binary values. A. Binarization", + "options": [ + "B. One-hot encoding", + "C. Tokenization", + "D. Normalization transformation" + ], + "correct": "B. One-hot encoding", + "explanation": "One-hot encoding is a technique that can be used to convert a categorical variable, such as the Day- Of_Week column, to binary values. One-hot encoding creates a new binary column for each unique value in the original column, and assigns a value o f 1 to the column that corresponds to the value in the original column, and 0 to the rest. For example , if the original column has values Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, and Sunday, one-hot encoding will create seven new columns, each representing one day of the week. If the value in the original column is Tuesday, then the column for Tuesday will have a value of 1, and the other columns will have a value of 0. One-hot encoding can help improve the performance o f machine learning models, as it eliminates the ordinal relationship between the values and cre ates a more informative and sparse representation of the data.", + "references": "One-Hot Encoding - Amazon SageMaker One-Hot Encoding: A Simple Guide for Beginners | by Jana Schmidt \u00a6 One-Hot Encoding in Machine Learning | by Nishant M alik | Towards \u00a6" + }, + { + "question": "A company has raw user and transaction data stored in AmazonS3 a MySQL database, and Amazon RedShift A Data Scientist needs to perform an analy sis by joining the three datasets from Amazon S3, MySQL, and Amazon RedShift, and then calculating th e average-of a few selected columns from the joined data Which AWS service should the Data Scientist use?", + "options": [ + "A. Amazon Athena", + "B. Amazon Redshift Spectrum", + "C. AWS Glue", + "D. Amazon QuickSight" + ], + "correct": "", + "explanation": "Amazon Athena is a serverless interactive query ser vice that can analyze data in Amazon S3 using standard SQL. Amazon Athena can also query data fro m other sources, such as MySQL and Amazon Redshift, by using federated queries. Federated que ries allow Amazon Athena to run SQL queries across data sources, such as relational and non-rel ational databases, data warehouses, and data lakes. By using Amazon Athena, the Data Scientist c an perform an analysis by joining the three datasets from Amazon S3, MySQL, and Amazon Redshift , and then calculating the average of a few selected columns from the joined data. Amazon Athen a can also integrate with other AWS services, such as AWS Glue and Amazon QuickSight, to provide additional features, such as data cataloging and visualization.", + "references": "What is Amazon Athena? - Amazon Athena Federated Query Overview - Amazon Athena Querying Data from Amazon S3 - Amazon Athena Querying Data from MySQL - Amazon Athena [Querying Data from Amazon Redshift - Amazon Athena ]" + }, + { + "question": "A Mobile Network Operator is building an analytics platform to analyze and optimize a company's operations using Amazon Athena and Amazon S3 The source systems send data in CSV format in real lime The Data Engineering team wants to transform the data to the Apache Parquet format bef ore storing it on Amazon S3 Which solution takes the LEAST effort to implement?", + "options": [ + "A. Ingest .CSV data using Apache Kafka Streams on Am azon EC2 instances and use Kafka Connect S3", + "B. Ingest .CSV data from Amazon Kinesis Data Streams and use Amazon Glue to convert data into", + "C. Ingest .CSV data using Apache Spark Structured St reaming in an Amazon EMR cluster and use", + "D. Ingest .CSV data from Amazon Kinesis Data Streams and use Amazon Kinesis Data Firehose to" + ], + "correct": "D. Ingest .CSV data from Amazon Kinesis Data Streams and use Amazon Kinesis Data Firehose to", + "explanation": "Amazon Kinesis Data Streams is a service that can c apture, store, and process streaming data in real time. Amazon Kinesis Data Firehose is a service tha t can deliver streaming data to various destinations, such as Amazon S3, Amazon Redshift, o r Amazon Elasticsearch Service. Amazon Kinesis Data Firehose can also transform the data before de livering it, such as converting the data format, compressing the data, or encrypting the data. One o f the supported data formats that Amazon Kinesis Data Firehose can convert to is Apache Parq uet, which is a columnar storage format that can improve the performance and cost-efficiency of anal ytics queries. By using Amazon Kinesis Data Streams and Amazon Kinesis Data Firehose, the Mobil e Network Operator can ingest the .CSV data from the source systems and use Amazon Kinesis Data Firehose to convert the data into Parquet before storing it on Amazon S3. This solution takes the least effort to implement, as it does not require any additional resources, such as Amazon EC 2 instances, Amazon EMR clusters, or Amazon Glue jobs. The solution can also leverage the built -in features of Amazon Kinesis Data Firehose, such as data buffering, batching, retry, and error handl ing. Reference: Amazon Kinesis Data Streams - Amazon Web Services Amazon Kinesis Data Firehose - Amazon Web Services Data Transformation - Amazon Kinesis Data Firehose Apache Parquet - Amazon Athena", + "references": "" + }, + { + "question": "An e-commerce company needs a customized training m odel to classify images of its shirts and pants products The company needs a proof of concept in 2 to 3 days with good accuracy Which compute choice should the Machine Learning Specialist selec t to train and achieve good accuracy on the model quickly?", + "options": [ + "A. m5 4xlarge (general purpose)", + "B. r5.2xlarge (memory optimized)", + "C. p3.2xlarge (GPU accelerated computing)", + "D. p3 8xlarge (GPU accelerated computing)" + ], + "correct": "C. p3.2xlarge (GPU accelerated computing)", + "explanation": "Image classification is a machine learning task tha t involves assigning labels to images based on thei r content. Image classification can be performed usin g various algorithms, such as convolutional neural networks (CNNs), which are a type of deep le arning model that can learn to extract high-level features from images. To train a customized image c lassification model, the e-commerce company needs a compute choice that can support the high co mputational demands of deep learning and provide good accuracy on the model quickly. A GPU a ccelerated computing instance, such as p3.2xlarge, is a suitable choice for this task, as it can leverage the parallel processing power of GP Us to speed up the training process and reduce the tra ining time. A p3.2xlarge instance has one NVIDIA Tesla V100 GPU, which can provide up to 125 teraflo ps of mixed-precision performance and 16 GB of GPU memory. A p3.2xlarge instance can also use vari ous deep learning frameworks, such as TensorFlow, PyTorch, MXNet, etc., to build and trai n the image classification model. A p3.2xlarge instance is also more cost-effective than a p3.8xla rge instance, which has four NVIDIA Tesla V100 GPUs, as the latter may not be necessary for a proo f of concept with a small dataset. Therefore, the Machine Learning Specialist should select p3.2xlarg e as the compute choice to train and achieve good accuracy on the model quickly.", + "references": "Amazon EC2 P3 Instances - Amazon Web Services Image Classification - Amazon SageMaker Convolutional Neural Networks - Amazon SageMaker Deep Learning AMIs - Amazon Web Services" + }, + { + "question": "A Marketing Manager at a pet insurance company plan s to launch a targeted marketing campaign on social media to acquire new customers Currently, th e company has the following data in Amazon Aurora Profiles for all past and existing customers Profiles for all past and existing insured pets Policy-level information Premiums received Claims paid What steps should be taken to implement a machine l earning model to identify potential new customers on social media?", + "options": [ + "A. Use regression on customer profile data to unders tand key characteristics of consumer segments", + "C. Use a recommendation engine on customer profile d ata to understand key characteristics of", + "D. Use a decision tree classifier engine on customer profile data to understand key characteristics of" + ], + "correct": "", + "explanation": "Clustering is a machine learning technique that can group data points into clusters based on their similarity or proximity. Clustering can help discov er the underlying structure and patterns in the dat a, as well as identify outliers or anomalies. Clusteri ng can also be used for customer segmentation, which is the process of dividing customers into gro ups based on their characteristics, behaviors, preferences, or needs. Customer segmentation can he lp understand the key features and needs of different customer segments, as well as design and implement targeted marketing campaigns for each segment. In this case, the Marketing Manager a t a pet insurance company plans to launch a targeted marketing campaign on social media to acqu ire new customers. To do this, the Manager can use clustering on customer profile data to understa nd the key characteristics of consumer segments, such as their demographics, pet types, policy prefe rences, premiums paid, claims made, etc. The Manager can then find similar profiles on social me dia, such as Facebook, Twitter, Instagram, etc., by using the cluster features as filters or keywords. The Manager can then target these potential new customers with personalized and relevant ads or off ers that match their segments needs and interests. This way, the Manager can implement a ma chine learning model to identify potential new customers on social media.", + "references": "" + }, + { + "question": "A company is running an Amazon SageMaker training j ob that will access data stored in its Amazon S3 bucket A compliance policy requires that the dat a never be transmitted across the internet How should the company set up the job?", + "options": [ + "A. Launch the notebook instances in a public subnet and access the data through the public S3", + "B. Launch the notebook instances in a private subnet and access the data through a NAT gateway", + "C. Launch the notebook instances in a public subnet and access the data through a NAT gateway", + "D. Launch the notebook instances in a private subnet and access the data through an S3 VPC" + ], + "correct": "D. Launch the notebook instances in a private subnet and access the data through an S3 VPC", + "explanation": "A private subnet is a subnet that does not have a r oute to the internet gateway, which means that the resources in the private subnet cannot access t he internet or be accessed from the internet. An S3 VPC endpoint is a gateway endpoint that allows t he resources in the VPC to access the S3 service without going through the internet. By launching th e notebook instances in a private subnet and accessing the data through an S3 VPC endpoint, the company can set up the job in a secure and compliant way, as the data never leaves the AWS net work and is not exposed to the internet. This can also improve the performance and reliability of the data transfer, as the traffic does not depend on the internet bandwidth or availability.", + "references": "Amazon VPC Endpoints - Amazon Virtual Private Cloud Endpoints for Amazon S3 - Amazon Virtual Private Cl oud Connect to SageMaker Within your VPC - Amazon SageM aker Working with VPCs and Subnets - Amazon Virtual Priv ate Cloud" + }, + { + "question": "A Machine Learning Specialist is preparing data for training on Amazon SageMaker The Specialist is transformed into a numpy .array, which appears to b e negatively affecting the speed of the training What should the Specialist do to optimize the data for training on SageMaker'?", + "options": [ + "A. Use the SageMaker batch transform feature to tran sform the training data into a DataFrame", + "B. Use AWS Glue to compress the data into the Apache Parquet format", + "C. Transform the dataset into the Recordio protobuf format", + "D. Use the SageMaker hyperparameter optimization fea ture to automatically optimize the data" + ], + "correct": "C. Transform the dataset into the Recordio protobuf format", + "explanation": "The Recordio protobuf format is a binary data forma t that is optimized for training on SageMaker. It allows faster data loading and lower memory usage c ompared to other formats such as CSV or numpy arrays. The Recordio protobuf format also sup ports features such as sparse input, variablelength input, and label embedding. To use the Recordio pro tobuf format, the data needs to be serialized and deserialized using the appropriate l ibraries. Some of the built-in algorithms in SageMaker support the Recordio protobuf format as a content type for training and inference. Reference: Common Data Formats for Training Using RecordIO Format Content Types Supported by Built-in Algorithms", + "references": "" + }, + { + "question": "A Machine Learning Specialist is training a model t o identify the make and model of vehicles in images The Specialist wants to use transfer learnin g and an existing model trained on images of general objects The Specialist collated a large cus tom dataset of pictures containing different vehicl e makes and models. What should the Specialist do to initialize the mod el to re-train it with the custom data?", + "options": [ + "A. Initialize the model with random weights in all l ayers including the last fully connected layer", + "B. Initialize the model with pre-trained weights in all layers and replace the last fully connected lay er.", + "C. Initialize the model with random weights in all l ayers and replace the last fully connected layer", + "D. Initialize the model with pre-trained weights in all layers including the last fully connected layer" + ], + "correct": "B. Initialize the model with pre-trained weights in all layers and replace the last fully connected lay er.", + "explanation": "Transfer learning is a technique that allows us to use a model trained for a certain task as a startin g point for a machine learning model for a different task. For image classification, a common practice i s to use a pre-trained model that was trained on a la rge and general dataset, such as ImageNet, and then customize it for the specific task. One way to customize the model is to replace the last fully connected layer, which is responsible for the final classification, with a new layer that has the same number of units as the number of classes in the new task. This way, the model can leverage the features learned by the previous layers, which are generic and useful for many image recognition tasks, and learn to map them to the new classes. Th e new layer can be initialized with random weights, and the rest of the model can be initializ ed with the pre-trained weights. This method is als o known as feature extraction, as it extracts meaning ful features from the pre-trained model and uses them for the new task. Reference: Transfer learning and fine-tuning Deep transfer learning for image classification: a survey", + "references": "" + }, + { + "question": "A Machine Learning Specialist is developing a custo m video recommendation model for an application The dataset used to train this model is very large with millions of data points and is hosted in an Amazon S3 bucket The Specialist wants to avoid loading all of this data onto an Amazon SageMaker notebook instance because it would take h ours to move and will exceed the attached 5 GB Amazon EBS volume on the notebook instance. Which approach allows the Specialist to use all the data to train the model?", + "options": [ + "A. Load a smaller subset of the data into the SageMa ker notebook and train locally. Confirm that the", + "B. Launch an Amazon EC2 instance with an AWS Deep Le arning AMI and attach the S3 bucket to the", + "C. Use AWS Glue to train a model using a small subse t of the data to confirm that the data will be", + "D. Load a smaller subset of the data into the SageMa ker notebook and train locally. Confirm that the" + ], + "correct": "A. Load a smaller subset of the data into the SageMa ker notebook and train locally. Confirm that the", + "explanation": "Pipe input mode is a feature of Amazon SageMaker th at allows streaming large datasets from Amazon S3 directly to the training algorithm withou t downloading them to the local disk. This reduces the startup time, disk space, and cost of t raining jobs. Pipe input mode is supported by most of the built-in algorithms and can also be used wit h custom training algorithms. To use Pipe input mode, the data needs to be in a binary format such as protobuf recordIO or TFRecord. The training code needs to use the PipeModeDataset class to read the data from the named pipe provided by SageMaker. To verify that the training code and the model parameters are working as expected, it is recommended to train locally on a smaller subset of the data before launching a full-scale training job on SageMaker. This approach is faster and more efficient than the other options, which involve either downloading the full dataset to an EC2 insta nce or using AWS Glue, which is not designed for training machine learning models. Reference: Using Pipe input mode for Amazon SageMaker algorith ms Using Pipe Mode with Your Own Algorithms PipeModeDataset Class", + "references": "" + }, + { + "question": "A Machine Learning Specialist is creating a new nat ural language processing application that processes a dataset comprised of 1 million sentence s The aim is to then run Word2Vec to generate embeddings of the sentences and enable different ty pes of predictions Here is an example from the dataset \"The quck BROWN FOX jumps over the lazy dog \" Which of the following are the operations the Speci alist needs to perform to correctly sanitize and prepare the data in a repeatable manner? (Select TH REE)", + "options": [ + "A. Perform part-of-speech tagging and keep the actio n verb and the nouns only", + "B. Normalize all words by making the sentence lowerc ase", + "C. Remove stop words using an English stopword dicti onary.", + "D. Correct the typography on \"quck\" to \"quick.\"" + ], + "correct": "", + "explanation": "To prepare the data for Word2Vec, the Specialist ne eds to perform some preprocessing steps that can help reduce the noise and complexity of the data, a s well as improve the quality of the embeddings. Some of the common preprocessing steps for Word2Vec are: Normalizing all words by making the sentence lowerc ase: This can help reduce the vocabulary size and treat words with different capitalizations as t he same word. For example, oeFox and oefox should be considered as the same word, not two different w ords. Removing stop words using an English stopword dicti onary: Stop words are words that are very common and do not carry much semantic meaning, such as oethe , oea , oeand , etc. Removing them can help focus on the words that are more relevant and informative for the task. Tokenizing the sentence into words: Tokenization is the process of splitting a sentence into smaller units, such as words or subwords. This is necessary for Word2Vec, as it operates on the word level and requires a list of words as input. The other options are not necessary or appropriate for Word2Vec: Performing part-of-speech tagging and keeping the a ction verb and the nouns only: Part-of-speech tagging is the process of assigning a grammatical c ategory to each word, such as noun, verb, adjective, etc. This can be useful for some natural language processing tasks, but not for Word2Vec, as it can lose some important information and conte xt by discarding other words. Correcting the typography on oequck to oequick : Ty po correction can be helpful for some tasks, but not for Word2Vec, as it can introduce errors and in consistencies in the data. For example, if the typo is intentional or part of a dialect, correcting it can change the meaning or style of the sentence. Moreover, Word2Vec can learn to handle typos and va riations in spelling by learning similar embeddings for them. One-hot encoding all words in the sentence: One-hot encoding is a way of representing words as vectors of 0s and 1s, where only one element is 1 a nd the rest are 0. The index of the 1 element corresponds to the words position in the vocabulary . For example, if the vocabulary is [oecat , oedog , oefox ], then oecat can be encoded as [1, 0, 0], oe dog as [0, 1, 0], and oefox as [0, 0, 1]. This can be useful for some machine learning models, but not fo r Word2Vec, as it does not capture the semantic similarity and relationship between words. Word2Vec aims to learn dense and low-dimensional embeddings for words, where similar words have simi lar vectors.", + "references": "" + }, + { + "question": "This graph shows the training and validation loss a gainst the epochs for a neural network The network being trained is as follows Two dense layers one output neuron 100 neurons in each layer 100 epochs Random initialization of weights Which technique can be used to improve model perfor mance in terms of accuracy in the validation set?", + "options": [ + "A. Early stopping", + "B. Random initialization of weights with appropriate seed", + "C. Increasing the number of epochs", + "D. Adding another layer with the 100 neurons" + ], + "correct": "A. Early stopping", + "explanation": "Early stopping is a technique that can be used to p revent overfitting and improve model performance on the validation set. Overfitting occu rs when the model learns the training data too well and fails to generalize to new and unseen data . This can be seen in the graph, where the training loss keeps decreasing, but the validation loss star ts to increase after some point. This means that th e model is fitting the noise and patterns in the trai ning data that are not relevant for the validation data. Early stopping is a way of stopping the train ing process before the model overfits the training data. It works by monitoring the validation loss an d stopping the training when the validation loss stops decreasing or starts increasing. This way, th e model is saved at the point where it has the best performance on the validation set. Early stopping c an also save time and resources by reducing the number of epochs needed for training. Reference: Early Stopping How to Stop Training Deep Neural Networks At the Ri ght Time Using Early Stopping", + "references": "" + }, + { + "question": "A manufacturing company asks its Machine Learning S pecialist to develop a model that classifies defective parts into one of eight defect types. The company has provided roughly 100000 images per defect type for training During the injial training of the image classification model the Specialist notices that the validation accuracy is 80%, while the training accuracy is 90% It is known that human-level performance for this type of image clas sification is around 90% What should the Specialist consider to fix this iss ue1?", + "options": [ + "A. A longer training time", + "B. Making the network larger", + "C. Using a different optimizer", + "D. Using some form of regularization" + ], + "correct": "D. Using some form of regularization", + "explanation": "Regularization is a technique that can be used to p revent overfitting and improve model performance on unseen data. Overfitting occurs when the model learns the training data too well and fails to generalize to new and unseen data. Thi s can be seen in the question, where the validation accuracy is lower than the training accu racy, and both are lower than the human-level performance. Regularization is a way of adding some constraints or penalties to the model to reduce its complexity and prevent it from memorizing the t raining data. Some common forms of regularization for image classification are: Weight decay: Adding a term to the loss function th at penalizes large weights in the model. This can help reduce the variance and noise in the model and make it more robust to small changes in the input. Dropout: Randomly dropping out some units or connec tions in the model during training. This can help reduce the co-dependency among the units and m ake the model more resilient to missing or corrupted features. Data augmentation: Artificially increasing the size and diversity of the training data by applying random transformations, such as cropping, flipping, rotating, scaling, etc. This can help the model learn more invariant and generalizable features and reduce the risk of overfitting to specific pattern s in the training data. The other options are not likely to fix the issue o f overfitting, and may even worsen it: A longer training time: This can lead to more overf itting, as the model will have more chances to fit the noise and details in the training data that are not relevant for the validation data. Making the network larger: This can increase the mo del capacity and complexity, which can also lead to more overfitting, as the model will have more pa rameters to learn and adjust to the training data. Using a different optimizer: This can affect the sp eed and stability of the training process, but not necessarily the generalization ability of the model . The choice of optimizer depends on the characteristics of the data and the model, and ther e is no guarantee that a different optimizer will prevent overfitting.", + "references": "Regularization (machine learning) Image Classification: Regularization How to Reduce Overfitting With Dropout Regularizati on in Keras" + }, + { + "question": "Example Corp has an annual sale event from October to December. The company has sequential sales data from the past 15 years and wants to use Amazon ML to predict the sales for this year's upcoming event. Which method should Example Corp use to split the d ata into a training dataset and evaluation dataset?", + "options": [ + "A. Pre-split the data before uploading to Amazon S3", + "B. Have Amazon ML split the data randomly.", + "C. Have Amazon ML split the data sequentially.", + "D. Perform custom cross-validation on the data Correct Answer: C" + ], + "correct": "", + "explanation": "A sequential split is a method of splitting data in to training and evaluation datasets while preservin g the order of the data records. This method is usefu l when the data has a temporal or sequential structure, and the order of the data matters for th e prediction task. For example, if the data contain s sales data for different months or years, and the g oal is to predict the sales for the next month or year, a sequential split can ensure that the traini ng data comes from the earlier period and the evaluation data comes from the later period. This c an help avoid data leakage, which occurs when the training data contains information from the fut ure that is not available at the time of prediction . A sequential split can also help evaluate the model performance on the most recent data, which may be more relevant and representative of the future d ata. In this question, Example Corp has sequential sales data from the past 15 years and wants to use Amazon ML to predict the sales for this years upcom ing annual sale event. A sequential split is the most appropriate method for splitting the data, as it can preserve the order of the data and prevent data leakage. For example, Example Corp can use the data from the first 14 years as the training dataset, and the data from the last year as the eva luation dataset. This way, the model can learn from the historical data and be tested on the most recen t data. Amazon ML provides an option to split the data sequ entially when creating the training and evaluation datasources. To use this option, Example Corp can specify the percentage of the data to use for training and evaluation, and Amazon ML will use the first part of the data for training and th e remaining part of the data for evaluation. For more information, see Splitting Your Data - Amazon Machine Learning.", + "references": "" + }, + { + "question": "A company is running a machine learning prediction service that generates 100 TB of predictions every day A Machine Learning Specialist must genera te a visualization of the daily precision-recall curve from the predictions, and forward a read-only version to the Business team. Which solution requires the LEAST coding effort?", + "options": [ + "A. Run a daily Amazon EMR workflow to generate preci sion-recall data, and save the results in", + "B. Generate daily precision-recall data in Amazon Qu ickSight, and publish the results in a dashboard", + "C. Run a daily Amazon EMR workflow to generate preci sion-recall data, and save the results in", + "D. Generate daily precision-recall data in Amazon ES , and publish the results in a dashboard shared" + ], + "correct": "C. Run a daily Amazon EMR workflow to generate preci sion-recall data, and save the results in", + "explanation": "A precision-recall curve is a plot that shows the t rade-off between the precision and recall of a binary classifier as the decision threshold is vari ed. It is a useful tool for evaluating and comparin g the performance of different models. To generate a precision-recall curve, the following steps are needed: Calculate the precision and recall values for diffe rent threshold values using the predictions and the true labels of the data. Plot the precision values on the y-axis and the rec all values on the x-axis for each threshold value. Optionally, calculate the area under the curve (AUC ) as a summary metric of the model performance. Among the four options, option C requires the least coding effort to generate and share a visualization of the daily precision-recall curve f rom the predictions. This option involves the following steps: Run a daily Amazon EMR workflow to generate precisi on-recall data: Amazon EMR is a service that allows running big data frameworks, such as Apache Spark, on a managed cluster of EC2 instances. Amazon EMR can handle large-scale data processing a nd analysis, such as calculating the precision and recall values for different threshold values fr om 100 TB of predictions. Amazon EMR supports various languages, such as Python, Scala, and R, fo r writing the code to perform the calculations. Amazon EMR also supports scheduling workflows using Apache Airflow or AWS Step Functions, which can automate the daily execution of the code. Save the results in Amazon S3: Amazon S3 is a servi ce that provides scalable, durable, and secure object storage. Amazon S3 can store the precision-r ecall data generated by Amazon EMR in a costeffecti ve and accessible way. Amazon S3 supports various data formats, such as CSV, JSON, or Parquet, for storing the data. Amazon S3 also integ rates with other AWS services, such as Amazon QuickSight, for further processing and visualizatio n of the data. Visualize the arrays in Amazon QuickSight: Amazon Q uickSight is a service that provides fast, easy-tou se, and interactive business intelligence and data visu alization. Amazon QuickSight can connect to Amazon S3 as a data source and import the precision -recall data into a dataset. Amazon QuickSight can then create a line chart to plot the precision- recall curve from the dataset. Amazon QuickSight also supports calculating the AUC and adding it as an annotation to the chart. Publish them in a dashboard shared with the Busines s team: Amazon QuickSight allows creating and publishing dashboards that contain one or more visu alizations from the datasets. Amazon QuickSight also allows sharing the dashboards with other users or groups within the same AWS account or across different AWS accounts. The Business team ca n access the dashboard with read-only permissions and view the daily precision-recall cur ve from the predictions. The other options require more coding effort than o ption C for the following reasons: Option A: This option requires writing code to plot the precision-recall curve from the data stored in Amazon S3, as well as creating a mechanism to share the plot with the Business team. This can involve using additional libraries or tools, such a s matplotlib, seaborn, or plotly, for creating the plot, and using email, web, or cloud services, such as AW S Lambda or Amazon SNS, for sharing the plot. Option B: This option requires transforming the pre dictions into a format that Amazon QuickSight can recognize and import as a data source, such as CSV, JSON, or Parquet. This can involve writing code to process and convert the predictions, as well as upl oading them to a storage service, such as Amazon S3 or Amazon Redshift, that Amazon QuickSight can c onnect to. Option D: This option requires writing code to gene rate precision-recall data in Amazon ES, as well as creating a dashboard to visualize the data. Amazon ES is a service that provides a fully managed Elasticsearch cluster, which is mainly used for sea rch and analytics purposes. Amazon ES is not designed for generating precision-recall data, and it requires using a specific data format, such as JSON, for storing the data. Amazon ES also requires using a tool, such as Kibana, for creating and sharing the dashboard, which can involve additional configuration and customization steps.", + "references": "Precision-Recall What Is Amazon EMR? What Is Amazon S3? [What Is Amazon QuickSight?] [What Is Amazon Elasticsearch Service?]" + }, + { + "question": "A Machine Learning Specialist has built a model usi ng Amazon SageMaker built-in algorithms and is not getting expected accurate results The Specialis t wants to use hyperparameter optimization to increase the model's accuracy Which method is the MOST repeatable and requires th e LEAST amount of effort to achieve this?", + "options": [ + "A. Launch multiple training jobs in parallel with di fferent hyperparameters", + "B. Create an AWS Step Functions workflow that monito rs the accuracy in Amazon CloudWatch Logs", + "C. Create a hyperparameter tuning job and set the ac curacy as an objective metric.", + "D. Create a random walk in the parameter space to it erate through a range of values that should be" + ], + "correct": "", + "explanation": "A hyperparameter tuning job is a feature of Amazon SageMaker that allows automatically finding the best combination of hyperparameters for a machine l earning model. Hyperparameters are high-level parameters that influence the learning process and the performance of the model, such as the learning rate, the number of layers, the regulariza tion factor, etc. A hyperparameter tuning job works by launching multiple training jobs with different hyperparameters, evaluating the results using an objective metric, and choosing the next set of hype rparameters to try based on a search strategy. The objective metric is a measure of the quality of the model, such as accuracy, precision, recall, etc. T he search strategy is a method of exploring the hyperp arameter space, such as random search, grid search, or Bayesian optimization. Among the four options, option C is the most repeat able and requires the least amount of effort to use hyperparameter optimization to increase the mod els accuracy. This option involves the following steps: Create a hyperparameter tuning job: Amazon SageMake r provides an easy-to-use interface for creating a hyperparameter tuning job, either throug h the AWS Management Console, the AWS CLI, or the AWS SDKs. To create a hyperparameter tuning job, the Machine Learning Specialist needs to specify the following information: The name and type of the algorithm to use, either a built-in algorithm or a custom algorithm. The ranges and types of the hyperparameters to tune , such as categorical, continuous, or integer. The name and type of the objective metric to optimi ze, such as accuracy, and whether to maximize or minimize it. The resource limits for the tuning job, such as the maximum number of training jobs and the maximum parallel training jobs. The input data channels and the output data locatio n for the training jobs. The configuration of the training instances, such a s the instance type, the instance count, the volume size, etc. Set the accuracy as an objective metric: To use acc uracy as an objective metric, the Machine Learning Specialist needs to ensure that the training algori thm writes the accuracy value to a file called metric_definitions in JSON format and prints it to stdout or stderr. For example, the file can contain the following content: This means that the training algorithm prints a lin e like this: Amazon SageMaker reads the accuracy value from the line and uses it to evaluate and compare the training jobs. The other options are not as repeatable and require more effort than option C for the following reasons: Option A: This option requires manually launching m ultiple training jobs in parallel with different hyperparameters, which can be tedious and error-pro ne. It also requires manually monitoring and comparing the results of the training jobs, which c an be time-consuming and subjective. Option B: This option requires writing code to crea te an AWS Step Functions workflow that monitors the accuracy in Amazon CloudWatch Logs and relaunch es the training job with a defined list of hyperparameters, which can be complex and challengi ng. It also requires maintaining and updating the list of hyperparameters, which can be inefficie nt and suboptimal. Option D: This option requires writing code to crea te a random walk in the parameter space to iterate through a range of values that should be us ed for each individual hyperparameter, which can be unreliable and unpredictable. It also requires d efining and implementing a stopping criterion, which can be arbitrary and inconsistent.", + "references": "Automatic Model Tuning - Amazon SageMaker Define Metrics to Monitor Model Performance" + }, + { + "question": "IT leadership wants Jo transition a company's exist ing machine learning data storage environment to AWS as a temporary ad hoc solution The company curr ently uses a custom software process that heavily leverages SOL as a query language and exclu sively stores generated csv documents for machine learning The ideal state for the company would be a solution that allows it to continue to use the current workforce of SQL experts The solution must also sup port the storage of csv and JSON files, and be able to query over semi-structured data The followi ng are high priorities for the company: Solution simplicity Fast development time Low cost High flexibility What technologies meet the company's requirements?", + "options": [ + "A. Amazon S3 and Amazon Athena", + "B. Amazon Redshift and AWS Glue", + "C. Amazon DynamoDB and DynamoDB Accelerator (DAX)", + "D. Amazon RDS and Amazon ES" + ], + "correct": "A. Amazon S3 and Amazon Athena", + "explanation": "Amazon S3 and Amazon Athena are technologies that m eet the companys requirements for a temporary ad hoc solution for machine learning data storage and query. Amazon S3 and Amazon Athena have the following features and benefits: Amazon S3 is a service that provides scalable, dura ble, and secure object storage for any type of data . Amazon S3 can store csv and JSON files, as well as other formats, and can handle large volumes of data with high availability and performance. Amazon S3 also integrates with other AWS services, such as Amazon Athena, for further processing and a nalysis of the data. Amazon Athena is a service that allows querying dat a stored in Amazon S3 using standard SQL. Amazon Athena can query over semi-structured data, such as JSON, as well as structured data, such as csv, without requiring any loading or transforma tion. Amazon Athena is serverless, meaning that there is no infrastructure to manage and users only pay for the queries they run. Amazon Athena also supports the use of AWS Glue Data Catalog, which is a centralized metadata repository that can store and manage the schema and partition information of the data in Amazon S3. Using Amazon S3 and Amazon Athena, the company can achieve the following high priorities: Solution simplicity: Amazon S3 and Amazon Athena ar e easy to use and require minimal configuration and maintenance. The company can simp ly upload the csv and JSON files to Amazon S3 and use Amazon Athena to query them using SQL. The company does not need to worry about provisioning, scaling, or managing any servers or c lusters. Fast development time: Amazon S3 and Amazon Athena can enable the company to quickly access and analyze the data without any data preparation o r loading. The company can use the existing workforce of SQL experts to write and run queries o n Amazon Athena and get results in seconds or minutes. Low cost: Amazon S3 and Amazon Athena are cost-effe ctive and offer pay-as-you-go pricing models. Amazon S3 charges based on the amount of storage us ed and the number of requests made. Amazon Athena charges based on the amount of data scanned by the queries. The company can also reduce the costs by using compression, encryption, and par titioning techniques to optimize the data storage and query performance. High flexibility: Amazon S3 and Amazon Athena are f lexible and can support various data types, formats, and sources. The company can store and que ry any type of data in Amazon S3, such as csv, JSON, Parquet, ORC, etc. The company can also query data from multiple sources in Amazon S3, such as data lakes, data warehouses, log files, etc. The other options are not as suitable as option A f or the companys requirements for the following reasons: Option B: Amazon Redshift and AWS Glue are technolo gies that can be used for data warehousing and data integration, but they are not ideal for a temporary ad hoc solution. Amazon Redshift is a service that provides a fully managed, petabyte-sca le data warehouse that can run complex analytical queries using SQL. AWS Glue is a service that provides a fully managed extract, transform, and load (ETL) service that can prepare and load da ta for analytics. However, using Amazon Redshift and AWS Glue would require more effort and cost tha n using Amazon S3 and Amazon Athena. The company would need to load the data from Amazon S3 to Amazon Redshift using AWS Glue, which can take time and incur additional charges. The com pany would also need to manage the capacity and performance of the Amazon Redshift cluster, whi ch can be complex and expensive. Option C: Amazon DynamoDB and DynamoDB Accelerator (DAX) are technologies that can be used for fast and scalable NoSQL database and caching, b ut they are not suitable for the companys data storage and query needs. Amazon DynamoDB is a servi ce that provides a fully managed, key-value and document database that can deliver single-digit millisecond performance at any scale. DynamoDB Accelerator (DAX) is a service that provid es a fully managed, in-memory cache for DynamoDB that can improve the read performance by u p to 10 times. However, using Amazon DynamoDB and DAX would not allow the company to con tinue to use SQL as a query language, as Amazon DynamoDB does not support SQL. The company w ould need to use the DynamoDB API or the AWS SDKs to access and query the data, which ca n require more coding and learning effort. The company would also need to transform the csv and JS ON files into DynamoDB items, which can involve additional processing and complexity. Option D: Amazon RDS and Amazon ES are technologies that can be used for relational database and search and analytics, but they are not optimal for the companys data storage and query scenario. Amazon RDS is a service that provides a fully manag ed, relational database that supports various database engines, such as MySQL, PostgreSQL, Oracle , etc. Amazon ES is a service that provides a fully managed, Elasticsearch cluster, which is main ly used for search and analytics purposes. However, using Amazon RDS and Amazon ES would not b e as simple and cost-effective as using Amazon S3 and Amazon Athena. The company would need to load the data from Amazon S3 to Amazon RDS, which can take time and incur additiona l charges. The company would also need to manage the capacity and performance of the Amazon R DS and Amazon ES clusters, which can be complex and expensive. Moreover, Amazon RDS and Ama zon ES are not designed to handle semistructured data, such as JSON, as well as Amazon S3 and Amazon Athena.", + "references": "Amazon S3 Amazon Athena Amazon Redshift AWS Glue Amazon DynamoDB [DynamoDB Accelerator (DAX)] [Amazon RDS] [Amazon ES]" + }, + { + "question": "A Machine Learning Specialist is working for a cred it card processing company and receives an unbalanced dataset containing credit card transacti ons. It contains 99,000 valid transactions and 1,000 fraudulent transactions The Specialist is ask ed to score a model that was run against the dataset The Specialist has been advised that identi fying valid transactions is equally as important as identifying fraudulent transactions What metric is BEST suited to score the model?", + "options": [ + "A. Precision B. Recall", + "C. Area Under the ROC Curve (AUC)", + "D. Root Mean Square Error (RMSE)" + ], + "correct": "C. Area Under the ROC Curve (AUC)", + "explanation": "Area Under the ROC Curve (AUC) is a metric that is best suited to score the model for the given scenario. AUC is a measure of the performance of a binary classifier, such as a model that predicts whether a credit card transaction is valid or fraud ulent. AUC is calculated based on the Receiver Operating Characteristic (ROC) curve, which is a pl ot that shows the trade-off between the true positive rate (TPR) and the false positive rate (FP R) of the classifier as the decision threshold is varied. The TPR, also known as recall or sensitivit y, is the proportion of actual positive cases (fraudulent transactions) that are correctly predic ted as positive by the classifier. The FPR, also known as the fall-out, is the proportion of actual negative cases (valid transactions) that are incorrectly predicted as positive by the classifier . The ROC curve illustrates how well the classifier can distinguish between the two classes, regardless of the class distribution or the error costs. A perfec t classifier would have a TPR of 1 and an FPR of 0 fo r all thresholds, resulting in a ROC curve that goe s from the bottom left to the top left and then to th e top right of the plot. A random classifier would have a TPR and an FPR that are equal for all thresh olds, resulting in a ROC curve that goes from the bottom left to the top right of the plot along the diagonal line. AUC is the area under the ROC curve, and it ranges from 0 to 1. A higher AUC indicates a better classifier, as it means that the classifier has a higher TPR and a lower FPR for all thresholds. AU C is a useful metric for imbalanced classification problems, such as the credit card transaction datas et, because it is insensitive to the class imbalanc e and the error costs. AUC can capture the overall pe rformance of the classifier across all possible scenarios, and it can be used to compare different classifiers based on their ROC curves. The other options are not as suitable as AUC for th e given scenario for the following reasons: Precision: Precision is the proportion of predicted positive cases (fraudulent transactions) that are actually positive. Precision is a useful metric whe n the cost of a false positive is high, such as in spam detection or medical diagnosis. However, precision is not a good metric for imbalanced classification problems, because it can be misleadingly high when the positive class is rare. For example, a classifier that predicts all transactions as valid would have a precision of 0, but a very high accura cy of 99%. Precision is also dependent on the decision threshold and the error costs, which may vary for different scenarios. Recall: Recall is the same as the TPR, and it is th e proportion of actual positive cases (fraudulent transactions) that are correctly predicted as posit ive by the classifier. Recall is a useful metric wh en the cost of a false negative is high, such as in fr aud detection or cancer diagnosis. However, recall is not a good metric for imbalanced classification pro blems, because it can be misleadingly low when the positive class is rare. For example, a classifi er that predicts all transactions as fraudulent wou ld have a recall of 1, but a very low accuracy of 1%. Recall is also dependent on the decision threshold and the error costs, which may vary for different s cenarios. Root Mean Square Error (RMSE): RMSE is a metric tha t measures the average difference between the predicted and the actual values. RMSE is a useful m etric for regression problems, where the goal is to predict a continuous value, such as the price of a house or the temperature of a city. However, RMSE is not a good metric for classification problems, w here the goal is to predict a discrete value, such as the class label of a transaction. RMSE is not meani ngful for classification problems, because it does not capture the accuracy or the error costs of the predictions.", + "references": "ROC Curve and AUC How and When to Use ROC Curves and Precision-Recall Curves for Classification in Python Precision-Recall Root Mean Squared Error" + }, + { + "question": "A bank's Machine Learning team is developing an app roach for credit card fraud detection The company has a large dataset of historical data labe led as fraudulent The goal is to build a model to take the information from new transactions and pred ict whether each transaction is fraudulent or not Which built-in Amazon SageMaker machine learning al gorithm should be used for modeling this problem?", + "options": [ + "A. Seq2seq", + "B. XGBoost", + "C. K-means", + "D. Random Cut Forest (RCF)" + ], + "correct": "B. XGBoost", + "explanation": "XGBoost is a built-in Amazon SageMaker machine lear ning algorithm that should be used for modeling the credit card fraud detection problem. X GBoost is an algorithm that implements a scalable and distributed gradient boosting framewor k, which is a popular and effective technique for supervised learning problems. Gradient boosting is a method of combining multiple weak learners, such as decision trees, into a strong learner, by i teratively fitting new models to the residual error s of the previous models and adding them to the ensemble . XGBoost can handle various types of data, such as numerical, categorical, or text, and can pe rform both regression and classification tasks. XGBoost also supports various features and optimiza tions, such as regularization, missing value handling, parallelization, and cross-validation, th at can improve the performance and efficiency of the algorithm. XGBoost is suitable for the credit card fraud detec tion problem for the following reasons: The problem is a binary classification problem, whe re the goal is to predict whether a transaction is fraudulent or not, based on the information from ne w transactions. XGBoost can perform binary classification by using a logistic regression objec tive function and outputting the probability of the positive class (fraudulent) for each transaction. The problem involves a large and imbalanced dataset of historical data labeled as fraudulent. XGBoost can handle large-scale and imbalanced data by using distributed and parallel computing, as well as techniques such as weighted sampling, class weighting, or stratified sampling, to balance the classes and reduce the bias towards the majority cl ass (non-fraudulent). The problem requires a high accuracy and precision for detecting fraudulent transactions, as well as a low false positive rate for avoiding false alarms. XGBoost can achieve high accuracy and precision by using gradient boosting, which can learn complex an d non-linear patterns from the data and reduce the variance and overfitting of the model. XGBoost can also achieve a low false positive rate by using regularization, which can reduce the complexity and noise of the model and prevent it from fitting spurious signals in the data. The other options are not as suitable as XGBoost fo r the credit card fraud detection problem for the following reasons: Seq2seq: Seq2seq is an algorithm that implements a sequence-to-sequence model, which is a type of neural network model that can map an input sequence to an output sequence. Seq2seq is mainly used for natural language processing tasks, such as machine translation, text summarization, or dialogue generation. Seq2seq is not suitable for th e credit card fraud detection problem, because the problem is not a sequence-to-sequence task, but a b inary classification task. The input and output of the problem are not sequences of words or tokens, b ut vectors of features and labels. K-means: K-means is an algorithm that implements a clustering technique, which is a type of unsupervised learning method that can group similar data points into clusters. K-means is mainly used for exploratory data analysis, dimensionality reduction, or anomaly detection. K-means is not suitable for the credit card fraud detection proble m, because the problem is not a clustering task, bu t a classification task. The problem requires using t he labeled data to train a model that can predict t he labels of new data, not finding the optimal number of clusters or the cluster memberships of the data. Random Cut Forest (RCF): RCF is an algorithm that i mplements an anomaly detection technique, which is a type of unsupervised learning method tha t can identify data points that deviate from the normal behavior or distribution of the data. RCF is mainly used for detecting outliers, frauds, or fau lts in the data. RCF is not suitable for the credit car d fraud detection problem, because the problem is not an anomaly detection task, but a classification task. The problem requires using the labeled data to train a model that can predict the labels of new data, not finding the anomaly scores or the anomalous data points in the data.", + "references": "XGBoost Algorithm Use XGBoost for Binary Classification with Amazon S ageMaker Seq2seq Algorithm K-means Algorithm [Random Cut Forest Algorithm]" + }, + { + "question": "While working on a neural network project, a Machin e Learning Specialist discovers thai some features in the data have very high magnitude resul ting in this data being weighted more in the cost function What should the Specialist do to ensure be tter convergence during backpropagation?", + "options": [ + "A. Dimensionality reduction", + "B. Data normalization", + "C. Model regulanzation", + "D. Data augmentation for the minority class" + ], + "correct": "B. Data normalization", + "explanation": "Data normalization is a data preprocessing techniqu e that scales the features to a common range, such as [0, 1] or [-1, 1]. This helps reduce the im pact of features with high magnitude on the cost function and improves the convergence during backpr opagation. Data normalization can be done using different methods, such as min-max scaling, z -score standardization, or unit vector normalization. Data normalization is different from dimensionality reduction, which reduces the number of features; model regularization, which add s a penalty term to the cost function to prevent overfitting; and data augmentation, which increases the amount of data by creating synthetic samples. Reference: Data processing options for AI/ML | AWS Machine Lea rning Blog Data preprocessing - Machine Learning Lens How to Normalize Data Using scikit-learn in Python Normalization | Machine Learning | Google for Devel opers", + "references": "" + }, + { + "question": "An online reseller has a large, multi-column datase t with one column missing 30% of its data A Machine Learning Specialist believes that certain c olumns in the dataset could be used to reconstruct the missing data. Which reconstruction approach should the Specialist use to preserve the integrity of the dataset?", + "options": [ + "A. Listwise deletion", + "B. Last observation carried forward", + "C. Multiple imputation", + "D. Mean substitution" + ], + "correct": "C. Multiple imputation", + "explanation": "Multiple imputation is a technique that uses machin e learning to generate multiple plausible values for each missing value in a dataset, based on the o bserved data and the relationships among the variables. Multiple imputation preserves the integr ity of the dataset by accounting for the uncertainty and variability of the missing data, an d avoids the bias and loss of information that may result from other methods, such as listwise deletio n, last observation carried forward, or mean substitution. Multiple imputation can improve the a ccuracy and validity of statistical analysis and machine learning models that use the imputed datase t. Reference: Managing missing values in your target and related datasets with automated imputation support in Amazon Forecast Imputation by feature importance (IBFI): A methodol ogy to impute missing data in large datasets Multiple Imputation by Chained Equations (MICE) Exp lained", + "references": "" + }, + { + "question": "A Machine Learning Specialist discover the followin g statistics while experimenting on a model. What can the Specialist from the experiments?", + "options": [ + "A. The model In Experiment 1 had a high variance err or lhat was reduced in Experiment 3 by", + "B. The model in Experiment 1 had a high bias error t hat was reduced in Experiment 3 by", + "C. The model in Experiment 1 had a high bias error a nd a high variance error that were reduced in", + "D. The model in Experiment 1 had a high random noise error that was reduced in Experiment 3 by" + ], + "correct": "A. The model In Experiment 1 had a high variance err or lhat was reduced in Experiment 3 by", + "explanation": "The model in Experiment 1 had a high variance error because it performed well on the training data (train error = 5%) but poorly on the test data (tes t error = 8%). This indicates that the model was overfitting the training data and not generalizing well to new data. The model in Experiment 3 had a lower variance error because it performed similarly on the training data (train error = 5.1%) and the test data (test error = 5.4%). This indicates that the model was more robust and less sensitive to the fluctuations in the training data. The model in Exp eriment 3 achieved this improvement by implementing regularization, which is a technique t hat reduces the complexity of the model and prevents overfitting by adding a penalty term to th e loss function. The model in Experiment 2 had a minimal bias error because it performed similarly o n the training data (train error = 5.2%) and the test data (test error = 5.7%) as the model in Exper iment 1. This indicates that the model was not underfitting the data and capturing the true relati onship between the input and output variables. The model in Experiment 2 increased the number of l ayers and neurons in the model, which is a way to increase the complexity and flexibility of the m odel. However, this did not improve the performance of the model, as the variance error rem ained high. This shows that increasing the complexity of the model is not always the best way to reduce the bias error, and may even increase the variance error if the model becomes too complex for the data. Reference: Bias Variance Tradeoff - Clearly Explained - Machin e Learning Plus The Bias-Variance Trade-off in Machine Learning - S tack Abuse", + "references": "" + }, + { + "question": "A Machine Learning Specialist needs to be able to i ngest streaming data and store it in Apache Parquet files for exploration and analysis. Which o f the following services would both ingest and store this data in the correct format?", + "options": [ + "A. AWSDMS", + "B. Amazon Kinesis Data Streams", + "C. Amazon Kinesis Data Firehose", + "D. Amazon Kinesis Data Analytics" + ], + "correct": "C. Amazon Kinesis Data Firehose", + "explanation": "Amazon Kinesis Data Firehose is a service that can ingest streaming data and store it in various destinations, including Amazon S3, Amazon Redshift, Amazon Elasticsearch Service, and Splunk. Amazon Kinesis Data Firehose can also convert the i ncoming data to Apache Parquet or Apache ORC format before storing it in Amazon S3. This can red uce the storage cost and improve the performance of analytical queries on the data. Amazon Kinesis D ata Firehose supports various data sources, such as Amazon Kinesis Data Streams, Amazon Managed Stre aming for Apache Kafka, AWS IoT, and custom applications. Amazon Kinesis Data Firehose c an also apply data transformation and compression using AWS Lambda functions. AWSDMS is not a valid service name. AWS Database Mi gration Service (AWS DMS) is a service that can migrate data from various sources to various ta rgets, but it does not support streaming data or Parquet format. Amazon Kinesis Data Streams is a service that can i ngest and process streaming data in real time, but it does not store the data in any destination. Amaz on Kinesis Data Streams can be integrated with Amazon Kinesis Data Firehose to store the data in P arquet format. Amazon Kinesis Data Analytics is a service that can analyze streaming data using SQL or Apache Flink, but it does not store the data in any destination. Amazon Kinesis Data Analytics can be integrated with Amazon Kinesis Data Firehose to store the data in Parquet format. Reference: Amazon Kinesis Data Firehose - Amazon Web Services What Is Amazon Kinesis Data Firehose? - Amazon Kine sis Data Firehose Amazon Kinesis Data Firehose FAQs - Amazon Web Serv ices", + "references": "" + }, + { + "question": "A Machine Learning Specialist needs to move and tra nsform data in preparation for training Some of the data needs to be processed in near-real time an d other data can be moved hourly There are existing Amazon EMR MapReduce jobs to clean and fea ture engineering to perform on the data Which of the following services can feed data to th e MapReduce jobs? (Select TWO )", + "options": [ + "A. AWSDMS", + "B. Amazon Kinesis", + "C. AWS Data Pipeline", + "D. Amazon Athena" + ], + "correct": "", + "explanation": "Explanation/Reference: Amazon Kinesis and AWS Data Pipeline are two servic es that can feed data to the Amazon EMR MapReduce jobs. Amazon Kinesis is a service that ca n ingest, process, and analyze streaming data in real time. Amazon Kinesis can be integrated with Am azon EMR to run MapReduce jobs on streaming data sources, such as web logs, social media, IoT d evices, and clickstreams. Amazon Kinesis can handle data that needs to be processed in near-real time, such as for anomaly detection, fraud detection, or dashboarding. AWS Data Pipeline is a service that can orchestrate and automate data movement and transformation across various AWS serv ices and on-premises data sources. AWS Data Pipeline can be integrated with Amazon EMR to run M apReduce jobs on batch data sources, such as Amazon S3, Amazon RDS, Amazon DynamoDB, and Amazon Redshift. AWS Data Pipeline can handle data that can be moved hourly, such as for data war ehousing, reporting, or machine learning. AWSDMS is not a valid service name. AWS Database Mi gration Service (AWS DMS) is a service that can migrate data from various sources to various ta rgets, but it does not support streaming data or MapReduce jobs. Amazon Athena is a service that can query data stor ed in Amazon S3 using standard SQL, but it does not feed data to Amazon EMR or run MapReduce jobs. Amazon ES is a service that provides a fully manage d Elasticsearch cluster, which can be used for search, analytics, and visualization, but it does n ot feed data to Amazon EMR or run MapReduce jobs. Reference: Using Amazon Kinesis with Amazon EMR - Amazon EMR AWS Data Pipeline - Amazon Web Services Using AWS Data Pipeline to Run Amazon EMR Jobs - AW S Data Pipeline", + "references": "" + }, + { + "question": "An insurance company is developing a new device for vehicles that uses a camera to observe drivers' behavior and alert them when they appear distracted The company created approximately 10,000 training images in a controlled environment that a Machine Learning Specialist will use to train and evaluate machine learning models During the model evaluation the Specialist notices that the training error rate diminishes faster as the number of epochs increases and the model is not accurately inferring on the unseen test images Which of the following should be used to resolve th is issue? (Select TWO)", + "options": [ + "A. Add vanishing gradient to the model", + "B. Perform data augmentation on the training data", + "C. Make the neural network architecture complex.", + "D. Use gradient checking in the model" + ], + "correct": "", + "explanation": "The issue described in the question is a sign of ov erfitting, which is a common problem in machine learning when the model learns the noise and detail s of the training data too well and fails to generalize to new and unseen data. Overfitting can result in a low training error rate but a high test error rate, which indicates poor performance and va lidity of the model. There are several techniques that can be used to prevent or reduce overfitting, such as data augmentation and regularization. Data augmentation is a technique that applies vario us transformations to the original training data, such as rotation, scaling, cropping, flipping, addi ng noise, changing brightness, etc., to create new and diverse data samples. Data augmentation can inc rease the size and diversity of the training data, which can help the model learn more features and pa tterns and reduce the variance of the model. Data augmentation is especially useful for image da ta, as it can simulate different scenarios and perspectives that the model may encounter in real l ife. For example, in the question, the device uses a camera to observe drivers behavior, so data augme ntation can help the model deal with different lighting conditions, angles, distances, etc. Data a ugmentation can be done using various libraries and frameworks, such as TensorFlow, PyTorch, Keras, Ope nCV, etc12 Regularization is a technique that adds a penalty t erm to the models objective function, which is typically based on the models parameters. Regulariz ation can reduce the complexity and flexibility of the model, which can prevent overfitting by avoi ding learning the noise and details of the training data. Regularization can also improve the stability and robustness of the model, as it can reduce the sensitivity of the model to small fluctuations in t he data. There are different types of regularizatio n, such as L1, L2, dropout, etc., but they all have th e same goal of reducing overfitting. L2 regularization, also known as weight decay or ridge regression, is one of the most common and effective regularization techniques. L2 regularizat ion adds the squared norm of the models parameters multiplied by a regularization parameter (lambda) to the models objective function. L2 regularization can shrink the models parameters tow ards zero, which can reduce the variance of the model and improve the generalization ability of the model. L2 regularization can be implemented using various libraries and frameworks, such as Ten sorFlow, PyTorch, Keras, Scikit-learn, etc34 The other options are not valid or relevant for res olving the issue of overfitting. Adding vanishing gradient to the model is not a technique, but a pro blem that occurs when the gradient of the models objective function becomes very small and the model stops learning. Making the neural network architecture complex is not a solution, but a possi ble cause of overfitting, as a complex model can have more parameters and more flexibility to fit th e training data too well. Using gradient checking i n the model is not a technique, but a debugging metho d that verifies the correctness of the gradient computation in the model. Gradient checking is not related to overfitting, but to the implementation of the model.", + "references": "" + }, + { + "question": "The Chief Editor for a product catalog wants the Re search and Development team to build a machine learning system that can be used to detect whether or not individuals in a collection of images are wearing the company's retail brand The team has a s et of training data Which machine learning algorithm should the researc hers use that BEST meets their requirements?", + "options": [ + "A. Latent Dirichlet Allocation (LDA)", + "B. Recurrent neural network (RNN)", + "C. K-means", + "D. Convolutional neural network (CNN)" + ], + "correct": "D. Convolutional neural network (CNN)", + "explanation": "A convolutional neural network (CNN) is a type of m achine learning algorithm that is suitable for image classification tasks. A CNN consists of multi ple layers that can extract features from images and learn to recognize patterns and objects. A CNN can also use transfer learning to leverage pretrain ed models that have been trained on large-scale image datasets, such as ImageNet, and finetune them for specific tasks, such as detecting the comp anys retail brand. A CNN can achieve high accuracy and performance for image classification p roblems, as it can handle complex and diverse images and reduce the dimensionality and noise of t he input data. A CNN can be implemented using various frameworks and libraries, such as TensorFlo w, PyTorch, Keras, MXNet, etc12 The other options are not valid or relevant for the image classification task. Latent Dirichlet Allocation (LDA) is a type of machine learning algo rithm that is suitable for topic modeling tasks. LD A can discover the hidden topics and their proportion s in a collection of text documents, such as news articles, tweets, reviews, etc. LDA is not applicab le for image data, as it requires textual input and output. LDA can be implemented using various framew orks and libraries, such as Gensim, Scikitlearn, Mallet, etc34 Recurrent neural network (RNN) is a type of machine learning algorithm that is suitable for sequential data tasks. RNN can process and generate data that has temporal or sequential dependencies, such as natural language, speech, aud io, video, etc. RNN is not optimal for image data, as it does not capture the spatial features a nd relationships of the pixels. RNN can be implemented using various frameworks and libraries, such as TensorFlow, PyTorch, Keras, MXNet, etc. K-means is a type of machine learning algorithm tha t is suitable for clustering tasks. K-means can partition a set of data points into a predefined nu mber of clusters, based on the similarity and distance between the data points. K-means is not su itable for image classification tasks, as it does not learn to label the images or detect the objects of interest. K-means can be implemented using various frameworks and libraries, such as Scikit-le arn, TensorFlow, PyTorch, etc.", + "references": "" + }, + { + "question": "A Machine Learning Specialist kicks off a hyperpara meter tuning job for a tree-based ensemble model using Amazon SageMaker with Area Under the RO C Curve (AUC) as the objective metric This workflow will eventually be deployed in a pipeline that retrains and tunes hyperparameters each night to model click-through on data that goes stal e every 24 hours With the goal of decreasing the amount of time it t akes to train these models, and ultimately to decrease costs, the Specialist wants to reconfigure the input hyperparameter range(s) Which visualization will accomplish this?", + "options": [ + "A. A histogram showing whether the most important in put feature is Gaussian.", + "B. A scatter plot with points colored by target vari able that uses (-Distributed Stochastic Neighbor", + "C. A scatter plot showing (he performance of the obj ective metric over each training iteration", + "D. A scatter plot showing the correlation between ma ximum tree depth and the objective metric." + ], + "correct": "D. A scatter plot showing the correlation between ma ximum tree depth and the objective metric.", + "explanation": "A scatter plot showing the correlation between maxi mum tree depth and the objective metric is a visualization that can help the Machine Learning Sp ecialist reconfigure the input hyperparameter range(s) for the tree-based ensemble model. A scatt er plot is a type of graph that displays the relationship between two variables using dots, wher e each dot represents one observation. A scatter plot can show the direction, strength, and shape of the correlation between the variables, as well as any outliers or clusters. In this case, the scatter plot can show how the maximum tree depth, which is a hyperparameter that controls the complexity and d epth of the decision trees in the ensemble model, affects the AUC, which is the objective metr ic that measures the performance of the model in terms of the trade-off between true positive rate a nd false positive rate. By looking at the scatter plot, the Machine Learning Specialist can see if th ere is a positive, negative, or no correlation between the maximum tree depth and the AUC, and how strong or weak the correlation is. The Machine Learning Specialist can also see if there i s an optimal value or range of values for the maximum tree depth that maximizes the AUC, or if th ere is a point of diminishing returns or overfitting where increasing the maximum tree depth does not improve or even worsens the AUC. Based on the scatter plot, the Machine Learning Spe cialist can reconfigure the input hyperparameter range(s) for the maximum tree depth to focus on the values that yield the best AUC, and avoid the values that result in poor AUC. This can decrease t he amount of time and cost it takes to train the model, as the hyperparameter tuning job can explore fewer and more promising combinations of values. A scatter plot can be created using various tools and libraries, such as Matplotlib, Seaborn, Plotly, etc12 The other options are not valid or relevant for rec onfiguring the input hyperparameter range(s) for the tree-based ensemble model. A histogram showing whether the most important input feature is Gaussian is a visualization that can help the Machi ne Learning Specialist understand the distribution and shape of the input data, but not the hyperparam eters. A histogram is a type of graph that displays the frequency or count of values in a sing le variable using bars, where each bar represents a bin or interval of values. A histogram can show if the variable is symmetric, skewed, or multimodal, and if it follows a normal or Gaussian distribution , which is a bell-shaped curve that is often assume d by many machine learning algorithms. In this case, the histogram can show if the most important input feature, which is a variable that has the mos t influence or predictive power on the output variable, is Gaussian or not. However, this does no t help the Machine Learning Specialist reconfigure the input hyperparameter range(s) for the tree-base d ensemble model, as the input feature is not a hyperparameter that can be tuned or optimized. A hi stogram can be created using various tools and libraries, such as Matplotlib, Seaborn, Plotly, etc 34 A scatter plot with points colored by target variab le that uses t-Distributed Stochastic Neighbor Embedding (t-SNE) to visualize the large number of input variables in an easier-to-read dimension is a visualization that can help the Machine Learning Specialist understand the structure and clustering of the input data, but not the hyperparameters. t-S NE is a technique that can reduce the dimensionality of high-dimensional data, such as im ages, text, or gene expression, and project it onto a lower-dimensional space, such as two or thre e dimensions, while preserving the local similarities and distances between the data points. t-SNE can help visualize and explore the patterns and relationships in the data, such as the clusters , outliers, or separability of the classes. In this case, the scatter plot can show how the input variables, which are the features or predictors of the output variable, are mapped onto a two-dimensional space u sing t-SNE, and how the points are colored by the target variable, which is the output or respons e variable that the model tries to predict. However , this does not help the Machine Learning Specialist reconfigure the input hyperparameter range(s) for the tree-based ensemble model, as the input variabl es and the target variable are not hyperparameters that can be tuned or optimized. A s catter plot with t-SNE can be created using various tools and libraries, such as Scikit-learn, TensorFlow, PyTorch, etc5 A scatter plot showing the performance of the objec tive metric over each training iteration is a visualization that can help the Machine Learning Sp ecialist understand the learning curve and convergence of the model, but not the hyperparamete rs. A scatter plot is a type of graph that displays the relationship between two variables usi ng dots, where each dot represents one observation. A scatter plot can show the direction, strength, and shape of the correlation between the variables, as well as any outliers or clusters. In this case, the scatter plot can show how the objective metric, which is the performance measure that the model tries to optimize, changes over each training iteration, which is the number of tim es that the model updates its parameters using a batch of data. A scatter plot can show if the objec tive metric improves, worsens, or stagnates over time, and if the model converges to a stable value or oscillates or diverges. However, this does not help the Machine Learning Specialist reconfigure th e input hyperparameter range(s) for the treebased ensemble model, as the objective metric and the tra ining iteration are not hyperparameters that can be tuned or optimized. A scatter plot can be created using various tools and libraries, such as Matplotlib, Seaborn, Plotly, etc.", + "references": "" + }, + { + "question": "A Machine Learning Specialist is configuring automa tic model tuning in Amazon SageMaker When using the hyperparameter optimization feature, which of the following guidelines should be followed to improve optimization? Choose the maximum number of hyperparameters suppor ted by", + "options": [ + "A. Amazon SageMaker to search the largest number of combinations possible", + "B. Specify a very large hyperparameter range to allo w Amazon SageMaker to cover every possible", + "C. Use log-scaled hyperparameters to allow the hyper parameter space to be searched as quickly as", + "D. Execute only one hyperparameter tuning job at a t ime and improve tuning through successive" + ], + "correct": "C. Use log-scaled hyperparameters to allow the hyper parameter space to be searched as quickly as", + "explanation": "Using log-scaled hyperparameters is a guideline tha t can improve the automatic model tuning in Amazon SageMaker. Log-scaled hyperparameters are hy perparameters that have values that span several orders of magnitude, such as learning rate, regularization parameter, or number of hidden units. Log-scaled hyperparameters can be specified by using a log-uniform distribution, which assigns equal probability to each order of magnitude within a range. For example, a log-uniform distribution between 0.001 and 1000 can sample values such as 0. 001, 0.01, 0.1, 1, 10, 100, or 1000 with equal probability. Using log-scaled hyperparameters can a llow the hyperparameter optimization feature to search the hyperparameter space more efficiently an d effectively, as it can explore different scales o f values and avoid sampling values that are too small or too large. Using log-scaled hyperparameters can also help avoid numerical issues, such as under flow or overflow, that may occur when using linear-scaled hyperparameters. Using log-scaled hyp erparameters can be done by setting the ScalingType parameter to Logarithmic when defining the hyperparameter ranges in Amazon SageMaker12 The other options are not valid or relevant guideli nes for improving the automatic model tuning in Amazon SageMaker. Choosing the maximum number of hy perparameters supported by Amazon SageMaker to search the largest number of combinati ons possible is not a good practice, as it can increase the time and cost of the tuning job and ma ke it harder to find the optimal values. Amazon SageMaker supports up to 20 hyperparameters for tun ing, but it is recommended to choose only the most important and influential hyperparameters for the model and algorithm, and use default or fixed values for the rest3 Specifying a very large hyperparameter range to allow Amazon SageMaker to cover every possible value is not a good practic e, as it can result in sampling values that are irrelevant or impractical for the model and algorit hm, and waste the tuning budget. It is recommended to specify a reasonable and realistic h yperparameter range based on the prior knowledge and experience of the model and algorithm , and use the results of the tuning job to refine the range if needed4 Executing only one hype rparameter tuning job at a time and improving tuning through successive rounds of experiments is not a good practice, as it can limit the exploratio n and exploitation of the hyperparameter space and ma ke the tuning process slower and less efficient. It is recommended to use parallelism and concurrency to run multiple training jobs simultaneously and leverage the Bayesian optimizati on algorithm that Amazon SageMaker uses to guide the search for the best hyperparameter values 5", + "references": "" + }, + { + "question": "A large mobile network operating company is buildin g a machine learning model to predict customers who are likely to unsubscribe from the se rvice. The company plans to offer an incentive for these customers as the cost of churn is far gre ater than the cost of the incentive. The model produces the following confusion matrix a fter evaluating on a test dataset of 100 customers: Based on the model evaluation results, why is this a viable model for production?", + "options": [ + "A. The model is 86% accurate and the cost incurred b y the company as a result of false negatives is", + "B. The precision of the model is 86%, which is less than the accuracy of the model.", + "C. The model is 86% accurate and the cost incurred b y the company as a result of false positives is", + "D. The precision of the model is 86%, which is great er than the accuracy of the model." + ], + "correct": "C. The model is 86% accurate and the cost incurred b y the company as a result of false positives is", + "explanation": "Based on the model evaluation results, this is a vi able model for production because the model is 86% accurate and the cost incurred by the company a s a result of false positives is less than the fals e negatives. The accuracy of the model is the proport ion of correct predictions out of the total predictions, which can be calculated by adding the true positives and true negatives and dividing by the total number of observations. In this case, the accuracy of the model is (10 + 76) / 100 = 0.86, which means that the model correctly predicted 86% of the customers churn status. The cost incurred by the company as a result of false positi ves and false negatives is the loss or damage that the company suffers when the model makes incorrect predictions. A false positive is when the model predicts that a customer will churn, but the custom er actually does not churn. A false negative is when the model predicts that a customer will not ch urn, but the customer actually churns. In this case, the cost of a false positive is the incentive that the company offers to the customer who is predicted to churn, which is a relatively low cost. The cost of a false negative is the revenue that t he company loses when the customer churns, which is a relatively high cost. Therefore, the cost of a false positive is less than the cost of a false neg ative, and the company would prefer to have more false positives than false negatives. The model has 10 false positives and 4 false negatives, which means that the companys cost is lower than if the m odel had more false negatives and fewer false positives.", + "references": "" + }, + { + "question": "A Machine Learning Specialist is designing a system for improving sales for a company. The objective is to use the large amount of information the compa ny has on users' behavior and product preferences to predict which products users would l ike based on the users' similarity to other users. What should the Specialist do to meet this objectiv e?", + "options": [ + "A. Build a content-based filtering recommendation en gine with Apache Spark ML on Amazon EMR.", + "B. Build a collaborative filtering recommendation en gine with Apache Spark ML on Amazon EMR.", + "C. Build a model-based filtering recommendation engi ne with Apache Spark ML on Amazon EMR.", + "D. Build a combinative filtering recommendation engi ne with Apache Spark ML on Amazon EMR." + ], + "correct": "B. Build a collaborative filtering recommendation en gine with Apache Spark ML on Amazon EMR.", + "explanation": "A collaborative filtering recommendation engine is a type of machine learning system that can improve sales for a company by using the large amou nt of information the company has on users behavior and product preferences to predict which p roducts users would like based on the users similarity to other users. A collaborative filterin g recommendation engine works by finding the users who have similar ratings or preferences for the pro ducts, and then recommending the products that the similar users have liked but the target user ha s not seen or rated. A collaborative filtering recommendation engine can leverage the collective w isdom of the users and discover the hidden patterns and associations among the products and th e users. A collaborative filtering recommendation engine can be implemented using Apac he Spark ML on Amazon EMR, which are two services that can handle large-scale data proce ssing and machine learning tasks. Apache Spark ML is a library that provides various tools and alg orithms for machine learning, such as classificatio n, regression, clustering, recommendation, etc. Apache Spark ML can run on Amazon EMR, which is a service that provides a managed cluster platform th at simplifies running big data frameworks, such as Apache Spark, on AWS. Apache Spark ML on Amazon EMR can build a collaborative filtering recommendation engine using the Alternating Least S quares (ALS) algorithm, which is a matrix factorization technique that can learn the latent f actors that represent the users and the products, and then use them to predict the ratings or prefere nces of the users for the products. Apache Spark ML on Amazon EMR can also support both explicit fee dback, such as ratings or reviews, and implicit feedback, such as views or clicks, for building a c ollaborative filtering recommendation engine12", + "references": "" + }, + { + "question": "A Data Engineer needs to build a model using a data set containing customer credit card information. How can the Data Engineer ensure the data remains e ncrypted and the credit card information is secure?", + "options": [ + "A. Use a custom encryption algorithm to encrypt the data and store the data on an Amazon", + "B. Use an IAM policy to encrypt the data on the Amaz on S3 bucket and Amazon Kinesis to", + "C. Use an Amazon SageMaker launch configuration to e ncrypt the data once it is copied to the SageMaker", + "D. Use AWS KMS to encrypt the data on Amazon S3 and Amazon SageMaker, and redact the credit" + ], + "correct": "D. Use AWS KMS to encrypt the data on Amazon S3 and Amazon SageMaker, and redact the credit", + "explanation": "AWS KMS is a service that provides encryption and k ey management for data stored in AWS services and applications. AWS KMS can generate and manage e ncryption keys that are used to encrypt and decrypt data at rest and in transit. AWS KMS can al so integrate with other AWS services, such as Amazon S3 and Amazon SageMaker, to enable encryptio n of data using the keys stored in AWS KMS. Amazon S3 is a service that provides object storage for data in the cloud. Amazon S3 can use AWS KMS to encrypt data at rest using server-side encry ption with AWS KMS-managed keys (SSE-KMS). Amazon SageMaker is a service that provides a platf orm for building, training, and deploying machine learning models. Amazon SageMaker can use A WS KMS to encrypt data at rest on the SageMaker instances and volumes, as well as data in transit between SageMaker and other AWS services. AWS Glue is a service that provides a ser verless data integration platform for data preparation and transformation. AWS Glue can use AW S KMS to encrypt data at rest on the Glue Data Catalog and Glue ETL jobs. AWS Glue can also use bu ilt-in or custom classifiers to identify and redact sensitive data, such as credit card numbers, from t he customer data1234 The other options are not valid or secure ways to e ncrypt the data and protect the credit card information. Using a custom encryption algorithm to encrypt the data and store the data on an Amazon SageMaker instance in a VPC is not a good pr actice, as custom encryption algorithms are not recommended for security and may have flaws or vuln erabilities. Using the SageMaker DeepAR algorithm to randomize the credit card numbers is n ot a good practice, as DeepAR is a forecasting algorithm that is not designed for data anonymizati on or encryption. Using an IAM policy to encrypt the data on the Amazon S3 bucket and Amazon Kinesis to automatically discard credit card numbers and insert fake credit card numbers is not a good p ractice, as IAM policies are not meant for data encryption, but for access control and authorizatio n. Amazon Kinesis is a service that provides realti me data streaming and processing, but it does not have the capability to automatically discard or insert data values. Using an Amazon SageMaker launc h configuration to encrypt the data once it is copied to the SageMaker instance in a VPC is not a good practice, as launch configurations are not meant for data encryption, but for specifying the i nstance type, security group, and user data for the SageMaker instance. Using the SageMaker principal c omponent analysis (PCA) algorithm to reduce the length of the credit card numbers is not a good practice, as PCA is a dimensionality reduction algorithm that is not designed for data anonymizati on or encryption.", + "references": "" + }, + { + "question": "A Machine Learning Specialist is using an Amazon Sa geMaker notebook instance in a private subnet of a corporate VPC. The ML Specialist has important data stored on the Amazon SageMaker notebook instance's Amazon EBS volume, and needs to take a s napshot of that EBS volume. However the ML Specialist cannot find the Amazon SageMaker noteboo k instance's EBS volume or Amazon EC2 instance within the VPC. Why is the ML Specialist not seeing the instance vi sible in the VPC?", + "options": [ + "A. Amazon SageMaker notebook instances are based on the EC2 instances within the customer", + "B. Amazon SageMaker notebook instances are based on the Amazon ECS service within customer", + "C. Amazon SageMaker notebook instances are based on EC2 instances running within AWS service" + ], + "correct": "C. Amazon SageMaker notebook instances are based on EC2 instances running within AWS service", + "explanation": "Amazon SageMaker notebook instances are fully manag ed environments that provide an integrated Jupyter notebook interface for data exploration, an alysis, and machine learning. Amazon SageMaker notebook instances are based on EC2 instances that run within AWS service accounts, not within customer accounts. This means that the ML Specialis t cannot find the Amazon SageMaker notebook instances EC2 instance or EBS volume within the VPC , as they are not visible or accessible to the customer. However, the ML Specialist can still take a snapshot of the EBS volume by using the Amazon SageMaker console or API. The ML Specialist can also use VPC interface endpoints to securely connect the Amazon SageMaker notebook inst ance to the resources within the VPC, such as Amazon S3 buckets, Amazon EFS file systems, or Amaz on RDS databases", + "references": "" + }, + { + "question": "A manufacturing company has structured and unstruct ured data stored in an Amazon S3 bucket. A Machine Learning Specialist wants to use SQL to run queries on this data. Which solution requires the LEAST effort to be able to query this data?", + "options": [ + "A. Use AWS Data Pipeline to transform the data and A mazon RDS to run queries.", + "B. Use AWS Glue to catalogue the data and Amazon Ath ena to run queries.", + "C. Use AWS Batch to run ETL on the data and Amazon A urora to run the queries.", + "D. Use AWS Lambda to transform the data and Amazon K inesis Data Analytics to run queries." + ], + "correct": "B. Use AWS Glue to catalogue the data and Amazon Ath ena to run queries.", + "explanation": "Using AWS Glue to catalogue the data and Amazon Ath ena to run queries is the solution that requires the least effort to be able to query the d ata stored in an Amazon S3 bucket using SQL. AWS Glue is a service that provides a serverless data i ntegration platform for data preparation and transformation. AWS Glue can automatically discover , crawl, and catalogue the data stored in various sources, such as Amazon S3, Amazon RDS, Amazon Reds hift, etc. AWS Glue can also use AWS KMS to encrypt the data at rest on the Glue Data Catalog a nd Glue ETL jobs. AWS Glue can handle both structured and unstructured data, and support vario us data formats, such as CSV, JSON, Parquet, etc. AWS Glue can also use built-in or custom class ifiers to identify and parse the data schema and format1 Amazon Athena is a service that provides an interactive query engine that can run SQL queries directly on data stored in Amazon S3. Amazo n Athena can integrate with AWS Glue to use the Glue Data Catalog as a central metadata reposit ory for the data sources and tables. Amazon Athena can also use AWS KMS to encrypt the data at rest on Amazon S3 and the query results. Amazon Athena can query both structured and unstruc tured data, and support various data formats, such as CSV, JSON, Parquet, etc. Amazon Athena can also use partitions and compression to optimize the query performance and reduce the query cost23 The other options are not valid or require more eff ort to query the data stored in an Amazon S3 bucket using SQL. Using AWS Data Pipeline to transf orm the data and Amazon RDS to run queries is not a good option, as it involves moving the data f rom Amazon S3 to Amazon RDS, which can incur additional time and cost. AWS Data Pipeline is a se rvice that can orchestrate and automate data movement and transformation across various AWS serv ices and on-premises data sources. AWS Data Pipeline can be integrated with Amazon EMR to run E TL jobs on the data stored in Amazon S3. Amazon RDS is a service that provides a managed rel ational database service that can run various database engines, such as MySQL, PostgreSQL, Oracle , etc. Amazon RDS can use AWS KMS to encrypt the data at rest and in transit. Amazon RDS can run SQL queries on the data stored in the database tables45 Using AWS Batch to run ETL on the data and Amazon Aurora to run the queries is not a good option, as it also involves moving the data from Am azon S3 to Amazon Aurora, which can incur additional time and cost. AWS Batch is a service th at can run batch computing workloads on AWS. AWS Batch can be integrated with AWS Lambda to trig ger ETL jobs on the data stored in Amazon S3. Amazon Aurora is a service that provides a compatib le and scalable relational database engine that can run MySQL or PostgreSQL. Amazon Aurora can use AWS KMS to encrypt the data at rest and in transit. Amazon Aurora can run SQL queries on the d ata stored in the database tables. Using AWS Lambda to transform the data and Amazon Kinesis Dat a Analytics to run queries is not a good option, as it is not suitable for querying data stored in A mazon S3 using SQL. AWS Lambda is a service that can run serverless functions on AWS. AWS Lambda can be integrated with Amazon S3 to trigger data transformation functions on the data stored in Amaz on S3. Amazon Kinesis Data Analytics is a service that can analyze streaming data using SQL or Apache Flink. Amazon Kinesis Data Analytics can be integrated with Amazon Kinesis Data Streams or Amaz on Kinesis Data Firehose to ingest streaming data sources, such as web logs, social media, IoT d evices, etc. Amazon Kinesis Data Analytics is not designed for querying data stored in Amazon S3 usin g SQL.", + "references": "" + }, + { + "question": "A Machine Learning Specialist receives customer dat a for an online shopping website. The data includes demographics, past visits, and locality in formation. The Specialist must develop a machine learning approach to identify the customer shopping patterns, preferences and trends to enhance the website for better service and smart recommenda tions. Which solution should the Specialist recommend?", + "options": [ + "A. Latent Dirichlet Allocation (LDA) for the given c ollection of discrete data to identify patterns in the", + "B. A neural network with a minimum of three layers a nd random initial weights to identify patterns", + "C. Collaborative filtering based on user interaction s and correlations to identify patterns in the", + "D. Random Cut Forest (RCF) over random subsamples to identify patterns in the customer database" + ], + "correct": "C. Collaborative filtering based on user interaction s and correlations to identify patterns in the", + "explanation": "Collaborative filtering is a machine learning techn ique that recommends products or services to users based on the ratings or preferences of other users. This technique is well-suited for identifyin g customer shopping patterns and preferences because it takes into account the interactions between users and products.", + "references": "" + }, + { + "question": "A Machine Learning Specialist is working with a lar ge company to leverage machine learning within its products. The company wants to group its custom ers into categories based on which customers will and will not churn within the next 6 months. T he company has labeled the data available to the Specialist. Which machine learning model type should the Specia list use to accomplish this task?", + "options": [ + "A. Linear regression", + "B. Classification", + "C. Clustering", + "D. Reinforcement learning" + ], + "correct": "B. Classification", + "explanation": "The goal of classification is to determine to which class or category a data point (customer in our case) belongs to. For classification problems, data scientists would use historical data with predefined target variables AKA labels (churner/non -churner) \" answers that need to be predicted \" to train an algorithm. With classification, busines ses can answer the following questions: Will this customer churn or not? Will a customer renew their subscription? Will a user downgrade a pricing plan? Are there any signs of unusual customer behavior?", + "references": "https://www.kdnuggets.com9/05/churn-pred iction-machine-learning.html" + }, + { + "question": "The displayed graph is from a foresting model for t esting a time series. Considering the graph only, which conclusion should a Machine Learning Specialist make about the behavior of the model?", + "options": [ + "A. The model predicts both the trend and the seasona lity well.", + "B. The model predicts the trend well, but not the se asonality.", + "C. The model predicts the seasonality well, but not the trend.", + "D. The model does not predict the trend or the seaso nality well." + ], + "correct": "D. The model does not predict the trend or the seaso nality well.", + "explanation": "", + "references": "" + }, + { + "question": "A company wants to classify user behavior as either fraudulent or normal. Based on internal research, a Machine Learning Specialist would like to build a binary classifier based on two features:age of account and transaction month. The class dis tribution for these features is illustrated in the figure provided. Based on this information which model would have th e HIGHEST accuracy?", + "options": [ + "A. Long short-term memory (LSTM) model with scaled e xponential linear unit (SELL))", + "B. Logistic regression", + "C. Support vector machine (SVM) with non-linear kern el", + "D. Single perceptron with tanh activation function" + ], + "correct": "C. Support vector machine (SVM) with non-linear kern el", + "explanation": "Based on the figure provided, the data is not linea rly separable. Therefore, a non-linear model such as SVM with a non-linear kernel would be the best c hoice. SVMs are particularly effective in highdimen sional spaces and are versatile in that they can be used f or both linear and non-linear data. Additionally, SVMs have a high level of accur acy and are less prone to overfitting1", + "references": "1: https://docs.aws.amazon.com/sagemaker /latest/dg/svm.html" + }, + { + "question": "A Machine Learning Specialist at a company sensitiv e to security is preparing a dataset for model training. The dataset is stored in Amazon S3 and co ntains Personally Identifiable Information (Pll). The dataset: * Must be accessible from a VPC only. * Must not traverse the public internet. How can these requirements be satisfied?", + "options": [ + "A. Create a VPC endpoint and apply a bucket access p olicy that restricts access to the given VPC", + "B. Create a VPC endpoint and apply a bucket access p olicy that allows access from the given VPC", + "C. Create a VPC endpoint and use Network Access Cont rol Lists (NACLs) to allow traffic between only", + "D. Create a VPC endpoint and use security groups to restrict access to the given VPC endpoint and an" + ], + "correct": "A. Create a VPC endpoint and apply a bucket access p olicy that restricts access to the given VPC", + "explanation": "A VPC endpoint is a logical device that enables pri vate connections between a VPC and supported AWS services. A VPC endpoint can be either a gatewa y endpoint or an interface endpoint. A gateway endpoint is a gateway that is a target for a specif ied route in the route table, used for traffic dest ined to a supported AWS service. An interface endpoint i s an elastic network interface with a private IP address that serves as an entry point for traffic d estined to a supported service1 In this case, the Machine Learning Specialist can c reate a gateway endpoint for Amazon S3, which is a supported service for gateway endpoints. A gateway endpoint for Amazon S3 enables the VPC to access Amazon S3 privately, without requiring an in ternet gateway, NAT device, VPN connection, or AWS Direct Connect connection. The traffic between the VPC and Amazon S3 does not leave the Amazon network2 To restrict access to the dataset stored in Amazon S3, the Machine Learning Specialist can apply a bucket access policy that allows access only from t he given VPC endpoint and the VPC. A bucket access policy is a resource-based policy that defin es who can access a bucket and what actions they can perform. A bucket access policy can use various conditions to control access, such as the source IP address, the source VPC, the source VPC endpoint , etc. In this case, the Machine Learning Specialist can use the aws:sourceVpce condition to specify the ID of the VPC endpoint, and the aws:sourceVpc condition to specify the ID of the VP C. This way, only the requests that originate from the VPC endpoint or the VPC can access the bucket t hat contains the dataset34 The other options are not valid or secure ways to s atisfy the requirements. Creating a VPC endpoint and applying a bucket access policy that allows acc ess from the given VPC endpoint and an Amazon EC2 instance is not a good option, as it does not r estrict access to the VPC. An Amazon EC2 instance i s a virtual server that runs in the AWS cloud. An Ama zon EC2 instance can have a public IP address or a private IP address, depending on the network config uration. Allowing access from an Amazon EC2 instance does not guarantee that the instance is in the same VPC as the VPC endpoint, and may expose the dataset to unauthorized access. Creating a VPC endpoint and using Network Access Control Lists (NACLs) to allow traffic between only the given VPC endpoint and an Amazon EC2 instance is not a good option, as it does not restr ict access to the VPC. NACLs are stateless firewall s that can control inbound and outbound traffic at th e subnet level. NACLs can use rules to allow or deny traffic based on the protocol, port, and sourc e or destination IP address. However, NACLs do not support VPC endpoints as a source or destinatio n, and cannot filter traffic based on the VPC endpoint ID or the VPC ID. Therefore, using NACLs d oes not guarantee that the traffic is from the VPC endpoint or the VPC, and may expose the dataset to unauthorized access. Creating a VPC endpoint and using security groups to restrict access to the given VPC endpoint and an Amazon EC2 instance is not a good option, as it does not restrict access t o the VPC. Security groups are stateful firewalls t hat can control inbound and outbound traffic at the ins tance level. Security groups can use rules to allow or deny traffic based on the protocol, port, and so urce or destination. However, security groups do not support VPC endpoints as a source or destinatio n, and cannot filter traffic based on the VPC endpoint ID or the VPC ID. Therefore, using securit y groups does not guarantee that the traffic is from the VPC endpoint or the VPC, and may expose th e dataset to unauthorized access.", + "references": "" + }, + { + "question": "An employee found a video clip with audio on a comp any's social media feed. The language used in the video is Spanish. English is the employee's fir st language, and they do not understand Spanish. The employee wants to do a sentiment analysis. What combination of services is the MOST efficient to accomplish the task?", + "options": [ + "A. Amazon Transcribe, Amazon Translate, and Amazon C omprehend", + "B. Amazon Transcribe, Amazon Comprehend, and Amazon SageMaker seq2seq", + "C. Amazon Transcribe, Amazon Translate, and Amazon S ageMaker Neural Topic Model (NTM)", + "D. Amazon Transcribe, Amazon Translate, and Amazon S ageMaker BlazingText" + ], + "correct": "A. Amazon Transcribe, Amazon Translate, and Amazon C omprehend", + "explanation": "Amazon Transcribe, Amazon Translate, and Amazon Com prehend are the most efficient combination of services to accomplish the task of sentiment ana lysis on a video clip with audio in Spanish. Amazon Transcribe is a service that can convert spe ech to text using deep learning. Amazon Transcribe can transcribe audio from various source s, such as video files, audio files, or streaming audio. Amazon Transcribe can also recognize multipl e speakers, different languages, accents, dialects, and custom vocabularies. In this case, Am azon Transcribe can transcribe the audio from the video clip in Spanish to text in Spanish1 Amazon Tr anslate is a service that can translate text from one language to another using neural machine transl ation. Amazon Translate can translate text from various sources, such as documents, web pages, chat messages, etc. Amazon Translate can also support multiple languages, domains, and styles. In this case, Amazon Translate can translate the text from Spanish to English2 Amazon Comprehend is a service that can analyze and derive insights from text using natural language processing. Amazon Comprehend can perform various tasks, such as sentiment analysis, entity recognition, key phra se extraction, topic modeling, etc. Amazon Comprehend can also support multiple languages and domains. In this case, Amazon Comprehend can perform sentiment analysis on the text in Engli sh and determine whether the feedback is positive, negative, neutral, or mixed3 The other options are not valid or efficient for ac complishing the task of sentiment analysis on a video clip with audio in Spanish. Amazon Comprehend , Amazon SageMaker seq2seq, and Amazon SageMaker Neural Topic Model (NTM) are not a good c ombination, as they do not include a service that can transcribe speech to text, which is a nece ssary step for processing the audio from the video clip. Amazon Comprehend, Amazon Translate, and Amaz on SageMaker BlazingText are not a good combination, as they do not include a service that can perform sentiment analysis, which is the main goal of the task. Amazon SageMaker BlazingText is a service that can train and deploy text classification and word embedding models using deep learning. Amazon SageMaker BlazingText can perform tasks such as text classification, named en tity recognition, part-of-speech tagging, etc., but not sentiment analysis4", + "references": "" + }, + { + "question": "A Machine Learning Specialist is packaging a custom ResNet model into a Docker container so the company can leverage Amazon SageMaker for training. The Specialist is using Amazon EC2 P3 instances to train the model and needs to properly configure the Docker container to leverage the NVIDIA GPUs. What does the Specialist need to do?", + "options": [ + "A. Bundle the NVIDIA drivers with the Docker image.", + "B. Build the Docker container to be NVIDIA-Docker co mpatible.", + "C. Organize the Docker container's file structure to execute on GPU instances.", + "D. Set the GPU flag in the Amazon SageMaker CreateTr ainingJob request body" + ], + "correct": "B. Build the Docker container to be NVIDIA-Docker co mpatible.", + "explanation": "To leverage the NVIDIA GPUs on Amazon EC2 P3 instan ces for training a custom ResNet model using Amazon SageMaker, the Machine Learning Specialist n eeds to build the Docker container to be NVIDIA-Docker compatible. NVIDIA-Docker is a tool t hat enables GPU-accelerated containers to run on Docker. NVIDIA-Docker can automatically configur e the Docker container with the necessary drivers, libraries, and environment variables to ac cess the NVIDIA GPUs. NVIDIA-Docker can also isolate the GPU resources and ensure that each cont ainer has exclusive access to a GPU. To build a Docker container that is NVIDIA-Docker c ompatible, the Machine Learning Specialist needs to follow these steps: Install the NVIDIA Container Toolkit on the host ma chine that runs Docker. This toolkit includes the NVIDIA Container Runtime, which is a modified versi on of the Docker runtime that supports GPU hardware. Use the base image provided by NVIDIA as the first line of the Dockerfile. The base image contains the NVIDIA drivers and CUDA toolkit that are requir ed for GPU-accelerated applications. The base image can be specified as FROM nvcr.io/nvidia/cuda: tag, where tag is the version of CUDA and the operating system. Install the required dependencies and frameworks fo r the ResNet model, such as PyTorch, torchvision, etc., in the Dockerfile. Copy the ResNet model code and any other necessary files to the Docker container in the Dockerfile. Build the Docker image using the docker build comma nd. Push the Docker image to a repository, such as Amaz on Elastic Container Registry (Amazon ECR), using the docker push command. Specify the Docker image URI and the instance type (ml.p3.xlarge) in the Amazon SageMaker CreateTrainingJob request body. The other options are not valid or sufficient for b uilding a Docker container that can leverage the NVIDIA GPUs on Amazon EC2 P3 instances. Bundling th e NVIDIA drivers with the Docker image is not a good option, as it can cause driver conflicts and compatibility issues with the host machine and the NVIDIA GPUs. Organizing the Docker containers file structure to execute on GPU instances is not a good option, as it does not ensure that the Docker container can access the NVIDIA GPUs and the CUDA toolkit. Setting the GPU flag in the Amazon Sa geMaker CreateTrainingJob request body is not a good option, as it does not apply to custom Docker containers, but only to built-in algorithms and frameworks that support GPU instances.", + "references": "" + }, + { + "question": "A Machine Learning Specialist is building a logisti c regression model that will predict whether or not a person will order a pizz", + "options": [ + "A. The Specialist is trying to build the optimal mod el with an ideal classification threshold.", + "B. Receiver operating characteristic (ROC) curve", + "C. Misclassification rate", + "D. Root Mean Square Error (RM&)" + ], + "correct": "A. The Specialist is trying to build the optimal mod el with an ideal classification threshold.", + "explanation": "A receiver operating characteristic (ROC) curve is a model evaluation technique that can be used to understand how different classification thresholds will impact the models performance. A ROC curve plots the true positive rate (TPR) against the fals e positive rate (FPR) for various values of the classification threshold. The TPR, also known as se nsitivity or recall, is the proportion of positive instances that are correctly classified as positive . The FPR, also known as the fall-out, is the proportion of negative instances that are incorrect ly classified as positive. A ROC curve can show the trade-off between the TPR and the FPR for different thresholds, and help the Machine Learning Specialist to select the optimal threshold that max imizes the TPR and minimizes the FPR. A ROC curve can also be used to compare the performance o f different models by calculating the area under the curve (AUC), which is a measure of how we ll the model can distinguish between the positive and negative classes. A higher AUC indicat es a better model", + "references": "" + }, + { + "question": "An interactive online dictionary wants to add a wid get that displays words used in similar contexts. A Machine Learning Specialist is asked to provide wor d features for the downstream nearest neighbor model powering the widget. What should the Specialist do to meet these require ments?", + "options": [ + "A. Create one-hot word encoding vectors.", + "B. Produce a set of synonyms for every word using Am azon Mechanical Turk.", + "C. Create word embedding factors that store edit dis tance with every other word.", + "D. Download word embeddings pre-trained on a large c orpus." + ], + "correct": "D. Download word embeddings pre-trained on a large c orpus.", + "explanation": "Word embeddings are a type of dense representation of words, which encode semantic meaning in a vector form. These embeddings are typically pre-tra ined on a large corpus of text data, such as a large set of books, news articles, or web pages, an d capture the context in which words are used. Word embeddings can be used as features for a neare st neighbor model, which can be used to find words used in similar contexts. Downloading pre-tra ined word embeddings is a good way to get started quickly and leverage the strengths of these representations, which have been optimized on a large amount of data. This is likely to result in m ore accurate and reliable features than other optio ns like one-hot encoding, edit distance, or using Amaz on Mechanical Turk to produce synonyms. Reference: https://aws.amazon.com/blogs/machine-lea rning/amazon-sagemaker-object2vec-addsnew- features-that-support-automatic-negative-sampling-a nd-speed-up-training/", + "references": "" + }, + { + "question": "A Machine Learning Specialist is configuring Amazon SageMaker so multiple Data Scientists can access notebooks, train models, and deploy endpoint s. To ensure the best operational performance, the Specialist needs to be able to track how often the Scientists are deploying models, GPU and CPU utilization on the deployed SageMaker endpoints, an d all errors that are generated when an endpoint is invoked. Which services are integrated with Amazon SageMaker to track this information? (Select TWO.)", + "options": [ + "A. AWS CloudTrail", + "B. AWS Health", + "C. AWS Trusted Advisor", + "D. Amazon CloudWatch" + ], + "correct": "", + "explanation": "The services that are integrated with Amazon SageMa ker to track the information that the Machine Learning Specialist needs are AWS CloudTrail and Am azon CloudWatch. AWS CloudTrail is a service that records the API calls and events for AWS servi ces, including Amazon SageMaker. AWS CloudTrail can track the actions performed by the Data Scienti sts, such as creating notebooks, training models, and deploying endpoints. AWS CloudTrail can also pr ovide information such as the identity of the user, the time of the action, the parameters used, and the response elements returned. AWS CloudTrail can help the Machine Learning Specialist to monitor the usage and activity of Amazon SageMaker, as well as to audit and troubleshoot any issues1 Amazon CloudWatch is a service that collects and analyzes the metrics and logs for AWS services, including Amazon SageMaker. Amazon CloudWatch can track the performance and utilizatio n of the Amazon SageMaker endpoints, such as the CPU and GPU utilization, the inference latency, the number of invocations, etc. Amazon CloudWatch can also track the errors and alarms tha t are generated when an endpoint is invoked, such as the model errors, the throttling errors, th e HTTP errors, etc. Amazon CloudWatch can help the Machine Learning Specialist to optimize the ope rational performance and reliability of Amazon SageMaker, as well as to set up notifications and a ctions based on the metrics and logs", + "references": "" + }, + { + "question": "A Machine Learning Specialist trained a regression model, but the first iteration needs optimizing. The Specialist needs to understand whether the mode l is more frequently overestimating or underestimating the target. What option can the Specialist use to determine whe ther it is overestimating or underestimating the target value?", + "options": [ + "A. Root Mean Square Error (RMSE)", + "B. Residual plots", + "C. Area under the curve", + "D. Confusion matrix" + ], + "correct": "B. Residual plots", + "explanation": "Residual plots are a model evaluation technique tha t can be used to understand whether a regression model is more frequently overestimating or underestimating the target. Residual plots are graphs that plot the residuals (the difference betw een the actual and predicted values) against the predicted values or other variables. Residual plots can help the Machine Learning Specialist to identify the patterns and trends in the residuals, such as the direction, shape, and distribution. Residual plots can also reveal the pr esence of outliers, heteroscedasticity, non-lineari ty, or other problems in the model12 To determine whether the model is overestimating or underestimating the target, the Machine Learning Specialist can use a residual plot that pl ots the residuals against the predicted values. Thi s type of residual plot is also known as a prediction error plot. A prediction error plot can show the magnitude and direction of the errors made by the m odel. If the model is overestimating the target, the residuals will be negative, and the points will be below the zero line. If the model is underestimating the target, the residuals will be p ositive, and the points will be above the zero line . If the model is accurate, the residuals will be clo se to zero, and the points will be scattered around the zero line. A prediction error plot can also sho w the variance and bias of the model. If the model has high variance, the residuals will have a large spread, and the points will be far from the zero li ne. If the model has high bias, the residuals will have a systematic pattern, such as a curve or a slope, and the points will not be randomly distributed aro und the zero line. A prediction error plot can help the Machine Learning Specialist to optimize the mod el by adjusting the complexity, features, or parameters of the model34 The other options are not valid or suitable for det ermining whether the model is overestimating or underestimating the target. Root Mean Square Error (RMSE) is a model evaluation metric that measures the average magnitude of the errors made b y the model. RMSE is the square root of the mean of the squared residuals. RMSE can indicate th e overall accuracy and performance of the model, but it cannot show the direction or distribu tion of the errors. RMSE can also be influenced by outliers or extreme values, and it may not be compa rable across different models or datasets5 Area under the curve (AUC) is a model evaluation metric that measures the ability of the model to distinguish between the positive and negative class es. AUC is the area under the receiver operating characteristic (ROC) curve, which plots the true po sitive rate against the false positive rate for var ious classification thresholds. AUC can indicate the ove rall quality and performance of the model, but it i s only applicable for binary classification models, n ot regression models. AUC cannot show the magnitude or direction of the errors made by the mo del. Confusion matrix is a model evaluation technique that summarizes the number of correct and incorrect predictions made by the model for each class. A confusion matrix is a table that show s the counts of true positives, false positives, tr ue negatives, and false negatives for each class. A co nfusion matrix can indicate the accuracy, precision , recall, and F1-score of the model for each class, b ut it is only applicable for classification models, not regression models. A confusion matrix cannot show t he magnitude or direction of the errors made by the model.", + "references": "" + }, + { + "question": "A company wants to classify user behavior as either fraudulent or normal. Based on internal research, a Machine Learning Specialist would like to build a binary classifier based on two features: age of account and transaction month. The class dis tribution for these features is illustrated in the figure provided. Based on this information, which model would have t he HIGHEST recall with respect to the fraudulent class?", + "options": [ + "A. Decision tree", + "B. Linear support vector machine (SVM)", + "C. Naive Bayesian classifier", + "D. Single Perceptron with sigmoidal activation funct ion" + ], + "correct": "A. Decision tree", + "explanation": "Based on the figure provided, a decision tree would have the highest recall with respect to the fraudulent class. Recall is a model evaluation metr ic that measures the proportion of actual positive instances that are correctly classified by the mode l. Recall is calculated as follows: Recall = True Positives / (True Positives + False N egatives) A decision tree is a type of machine learning model that can perform classification tasks by splitting the data into smaller and purer subsets based on a series of rules or conditions. A decision tree can handle both linear and non-linear data, and can cap ture complex patterns and interactions among the features. A decision tree can also be easily vi sualized and interpreted1 In this case, the data is not linearly separable, a nd has a clear pattern of seasonality. The fraudule nt class forms a large circle in the center of the plo t, while the normal class is scattered around the edges. A decision tree can use the transaction mont h and the age of account as the splitting criteria, and create a circular boundary that separates the f raudulent class from the normal class. A decision tree can achieve a high recall for the fraudulent c lass, as it can correctly identify most of the blac k dots as positive instances, and minimize the number of false negatives. A decision tree can also adjust the depth and complexity of the tree to bala nce the trade-off between recall and precision23 The other options are not valid or suitable for ach ieving a high recall for the fraudulent class. A li near support vector machine (SVM) is a type of machine l earning model that can perform classification tasks by finding a linear hyperplane that maximizes the margin between the classes. A linear SVM can handle linearly separable data, but not non-lin ear data. A linear SVM cannot capture the circular pattern of the fraudulent class, and may misclassif y many of the black dots as negative instances, resulting in a low recall4 A naive Bayesian classif ier is a type of machine learning model that can perform classification tasks by applying the Bayes theorem and assuming conditional independence among the features. A naive Bayesian classifier can handle both linear and non-linear data, and can incorporate prior knowledge and probabilities into the model. However, a naive Bayesian classifier may not perform well when the features are correlat ed or dependent, as in this case. A naive Bayesian classifier may not capture the circular pa ttern of the fraudulent class, and may misclassify many of the black dots as negative instances, resul ting in a low recall5 A single perceptron with sigmoidal activation function is a type of machine learning model that can perform classification tasks by applying a weighted linear combination of the features and a non-linear activation function. A single perceptron with sigmoidal activation funct ion can handle linearly separable data, but not non-linear data. A single perceptron with sigmoidal activation function cannot capture the circular pattern of the fraudulent class, and may misclassif y many of the black dots as negative instances, resulting in a low recall.", + "references": "" + }, + { + "question": "When submitting Amazon SageMaker training jobs usin g one of the built-in algorithms, which common parameters MUST be specified? (Select THREE. )", + "options": [ + "A. The training channel identifying the location of training data on an Amazon S3 bucket.", + "B. The validation channel identifying the location o f validation data on an Amazon S3 bucket.", + "C. The 1AM role that Amazon SageMaker can assume to perform tasks on behalf of the users.", + "D. Hyperparameters in a JSON array as documented for the algorithm used." + ], + "correct": "", + "explanation": "When submitting Amazon SageMaker training jobs usin g one of the built-in algorithms, the common parameters that must be specified are: The training channel identifying the location of tr aining data on an Amazon S3 bucket. This parameter tells SageMaker where to find the input d ata for the algorithm and what format it is in. For example, TrainingInputMode: File means that the input data is in files stored in S3. The IAM role that Amazon SageMaker can assume to pe rform tasks on behalf of the users. This parameter grants SageMaker the necessary permission s to access the S3 buckets, ECR repositories, and other AWS resources needed for the training job . For example, RoleArn: arn:aws:iam::123456789012:role/service-role/AmazonS ageMaker-ExecutionRole- 20200303T150948 means that SageMaker will use the s pecified role to run the training job. The output path specifying where on an Amazon S3 bu cket the trained model will persist. This parameter tells SageMaker where to save the model a rtifacts, such as the model weights and parameters, after the training job is completed. Fo r example, OutputDataConfig: {S3OutputPath: s3://my-bucket/my-training-job} means that SageMake r will store the model artifacts in the specified S3 location. The validation channel identifying the location of validation data on an Amazon S3 bucket is an optional parameter that can be used to provide a se parate dataset for evaluating the model performance during the training process. This param eter is not required for all algorithms and can be omitted if the validation data is not available or not needed. The hyperparameters in a JSON array as documented f or the algorithm used is another optional parameter that can be used to customize the behavio r and performance of the algorithm. This parameter is specific to each algorithm and can be used to tune the model accuracy, speed, complexity, and other aspects. For example, HyperPa rameters: {num_round: \"10\", objective: \"binary:logistic\"} means that the XGBoost algorithm will use 10 boosting rounds and the logistic loss function for binary classification. The Amazon EC2 instance class specifying whether tr aining will be run using CPU or GPU is not a parameter that is specified when submitting a train ing job using a built-in algorithm. Instead, this parameter is specified when creating a training ins tance, which is a containerized environment that runs the training code and algorithm. For example, ResourceConfig: {InstanceType: ml.m5.xlarge, InstanceCount: 1, VolumeSizeInGB: 10} means that Sa geMaker will use one m5.xlarge instance with 10 GB of storage for the training instance.", + "references": "Train a Model with Amazon SageMaker Use Amazon SageMaker Built-in Algorithms or Pre-tra ined Models CreateTrainingJob - Amazon SageMaker Service" + }, + { + "question": "A Data Scientist is developing a machine learning m odel to predict future patient outcomes based on information collected about each patient and their treatment plans. The model should output a continuous value as its prediction. The data availa ble includes labeled outcomes for a set of 4,000 patients. The study was conducted on a group of ind ividuals over the age of 65 who have a particular disease that is known to worsen with age. Initial models have performed poorly. While reviewi ng the underlying data, the Data Scientist notices that, out of 4,000 patient observations, there are 450 where the patient age has been input as 0. The other features for these observations appear normal compared to the rest of the sample population. How should the Data Scientist correct this issue?", + "options": [ + "A. Drop all records from the dataset where age has b een set to 0.", + "B. Replace the age field value for records with a va lue of 0 with the mean or median value from the", + "C. Drop the age feature from the dataset and train t he model using the rest of the features.", + "D. Use k-means clustering to handle missing features ." + ], + "correct": "B. Replace the age field value for records with a va lue of 0 with the mean or median value from the", + "explanation": "The best way to handle the missing values in the pa tient age feature is to replace them with the mean or median value from the dataset. This is a co mmon technique for imputing missing values that preserves the overall distribution of the data and avoids introducing bias or reducing the sample size. Dropping the records or the feature would res ult in losing valuable information and reducing the accuracy of the model. Using k-means clustering wou ld not be appropriate for handling missing values in a single feature, as it is a method for g rouping similar data points based on multiple features. Reference: Effective Strategies to Handle Missing Values in Da ta Analysis How To Handle Missing Values In Machine Learning Da ta With Weka How to handle missing values in Python - Machine Le arning Plus", + "references": "" + }, + { + "question": "A Data Science team is designing a dataset reposito ry where it will store a large amount of training data commonly used in its machine learning models. As Data Scientists may create an arbitrary number of new datasets every day the solution has t o scale automatically and be cost-effective. Also, it must be possible to explore the data using SQL. Which storage scheme is MOST adapted to this scenar io?", + "options": [ + "A. Store datasets as files in Amazon S3.", + "B. Store datasets as files in an Amazon EBS volume a ttached to an Amazon EC2 instance.", + "C. Store datasets as tables in a multi-node Amazon R edshift cluster.", + "D. Store datasets as global tables in Amazon DynamoD B." + ], + "correct": "A. Store datasets as files in Amazon S3.", + "explanation": "The best storage scheme for this scenario is to sto re datasets as files in Amazon S3. Amazon S3 is a scalable, cost-effective, and durable object storag e service that can store any amount and type of data. Amazon S3 also supports querying data using S QL with Amazon Athena, a serverless interactive query service that can analyze data directly in S3. This way, the Data Science team can easily explore and analyze their datasets without having to load t hem into a database or a compute instance. The other options are not as suitable for this scen ario because: Storing datasets as files in an Amazon EBS volume a ttached to an Amazon EC2 instance would limit the scalability and availability of the data, as EB S volumes are only accessible within a single availability zone and have a maximum size of 16 TiB . Also, EBS volumes are more expensive than S3 buckets and require provisioning and managing EC2 i nstances. Storing datasets as tables in a multi-node Amazon R edshift cluster would incur higher costs and complexity than using S3 and Athena. Amazon Redshif t is a data warehouse service that is optimized for analytical queries over structured or semi-stru ctured data. However, it requires setting up and maintaining a cluster of nodes, loading data into t ables, and choosing the right distribution and sort keys for optimal performance. Moreover, Amazon Reds hift charges for both storage and compute, while S3 and Athena only charge for the amount of d ata stored and scanned, respectively. Storing datasets as global tables in Amazon DynamoD B would not be feasible for large amounts of data, as DynamoDB is a key-value and document datab ase service that is designed for fast and consistent performance at any scale. However, Dynam oDB has a limit of 400 KB per item and 25 GB per partition key value, which may not be enough fo r storing large datasets. Also, DynamoDB does not support SQL queries natively, and would require using a service like Amazon EMR or AWS Glue to run SQL queries over DynamoDB data.", + "references": "Amazon S3 - Cloud Object Storage Amazon Athena \" Interactive SQL Queries for Data in Amazon S3 Amazon EBS - Amazon Elastic Block Store (EBS) Amazon Redshift \" Data Warehouse Solution - AWS Amazon DynamoDB \" NoSQL Cloud Database Service" + }, + { + "question": "A Machine Learning Specialist working for an online fashion company wants to build a data ingestion solution for the company's Amazon S3-based data lak e. The Specialist wants to create a set of ingestion m echanisms that will enable future capabilities comprised of: Real-time analytics Interactive analytics of historical data Clickstream analytics Product recommendations Which services should the Specialist use?", + "options": [ + "A. AWS Glue as the data dialog; Amazon Kinesis Data Streams and Amazon Kinesis Data Analytics for", + "B. Amazon Athena as the data catalog; Amazon Kinesis Data Streams and Amazon Kinesis Data", + "C. AWS Glue as the data catalog; Amazon Kinesis Data Streams and Amazon Kinesis Data Analytics for", + "D. Amazon Athena as the data catalog; Amazon Kinesis Data Streams and Amazon Kinesis Data", + "A. Customize the built-in image classification algor ithm to use Inception and use this for model", + "B. Create a support case with the SageMaker team to change the default image classification", + "C. Bundle a Docker container with TensorFlow Estimat or loaded with an Inception network and use", + "D. Use custom code in Amazon SageMaker with TensorFl ow Estimator to load the model with an" + ], + "correct": "A. AWS Glue as the data dialog; Amazon Kinesis Data Streams and Amazon Kinesis Data Analytics for", + "explanation": "The best options to use an Inception neural network architecture instead of a ResNet architecture for image classification in Amazon SageMaker are: Bundle a Docker container with TensorFlow Estimator loaded with an Inception network and use this for model training. This option allows users to cus tomize the training environment and use any TensorFlow model they want. Users can create a Dock er image that contains the TensorFlow Estimator API and the Inception model from the Tens orFlow Hub, and push it to Amazon ECR. Then, users can use the SageMaker Estimator class to trai n the model using the custom Docker image and the training data from Amazon S3. Use custom code in Amazon SageMaker with TensorFlow Estimator to load the model with an Inception network and use this for model training. This option allows users to use the built-in TensorFlow container provided by SageMaker and writ e custom code to load and train the Inception model. Users can use the TensorFlow Estimator class to specify the custom code and the training data from Amazon S3. The custom code can use the Te nsorFlow Hub module to load the Inception model and fine-tune it on the training data. The other options are not feasible for this scenari o because: Customize the built-in image classification algorit hm to use Inception and use this for model training . This option is not possible because the built-in im age classification algorithm in SageMaker does not support customizing the neural network architecture . The built-in algorithm only supports ResNet models with different depths and widths. Create a support case with the SageMaker team to ch ange the default image classification algorithm to Inception. This option is not realistic because the SageMaker team does not provide such a service. Users cannot request the SageMaker team to change the default algorithm or add new algorithms to the built-in ones. Download and apt-get install the inception network code into an Amazon EC2 instance and use this instance as a Jupyter notebook in Amazon SageMaker. This option is not advisable because it does not leverage the benefits of SageMaker, such as man aged training and deployment, distributed training, and automatic model tuning. Users would h ave to manually install and configure the Inception network code and the TensorFlow framework on the EC2 instance, and run the training and inference code on the same instance, which may not be optimal for performance and scalability.", + "references": "Use Your Own Algorithms or Models with Amazon SageM aker Use the SageMaker TensorFlow Serving Container TensorFlow Hub" + }, + { + "question": "Specialist ran into an overfitting problem in which the training and testing accuracies were 99% and 75%r respectively. How should the Specialist address this issue and wh at is the reason behind it?", + "options": [ + "A. The learning rate should be increased because the optimization process was trapped at a local", + "B. The dropout rate at the flatten layer should be i ncreased because the model is not generalized", + "C. The dimensionality of dense layer next to the fla tten layer should be increased because the model", + "D. The epoch number should be increased because the optimization process was terminated before" + ], + "correct": "B. The dropout rate at the flatten layer should be i ncreased because the model is not generalized", + "explanation": "The best way to address the overfitting problem in image classification is to increase the dropout rate at the flatten layer because the model is not generalized enough. Dropout is a regularization technique that randomly drops out some units from t he neural network during training, reducing the co-adaptation of features and preventing overfittin g. The flatten layer is the layer that converts the output of the convolutional layers into a one-dimen sional vector that can be fed into the dense layers. Increasing the dropout rate at the flatten layer means that more features from the convolutional layers will be ignored, forcing the m odel to learn more robust and generalizable representations from the remaining features. The other options are not correct for this scenario because: Increasing the learning rate would not help with th e overfitting problem, as it would make the optimization process more unstable and prone to ove rshooting the global minimum. A high learning rate can also cause the model to diverge or oscilla te around the optimal solution, resulting in poor performance and accuracy. Increasing the dimensionality of the dense layer ne xt to the flatten layer would not help with the overfitting problem, as it would make the model mor e complex and increase the number of parameters to be learned. A more complex model can fit the training data better, but it can also memorize the noise and irrelevant details in the da ta, leading to overfitting and poor generalization. Increasing the epoch number would not help with the overfitting problem, as it would make the model train longer and more likely to overfit the t raining data. A high epoch number can cause the model to converge to the global minimum, but it can also cause the model to over-optimize the training data and lose the ability to generalize to new data.", + "references": "Dropout: A Simple Way to Prevent Neural Networks fr om Overfitting How to Reduce Overfitting With Dropout Regularizati on in Keras How to Control the Stability of Training Neural Net works With the Learning Rate How to Choose the Number of Hidden Layers and Nodes in a Feedforward Neural Network? How to decide the optimal number of epochs to train a neural network?" + }, + { + "question": "A Machine Learning team uses Amazon SageMaker to tr ain an Apache MXNet handwritten digit classifier model using a research dataset. The team wants to receive a notification when the model is overfitting. Auditors want to view the Amazon SageM aker log activity report to ensure there are no unauthorized API calls. What should the Machine Learning team do to address the requirements with the least amount of code and fewest steps?", + "options": [ + "A. Implement an AWS Lambda function to long Amazon S ageMaker API calls to Amazon S3. Add code", + "C. Implement an AWS Lambda function to log Amazon Sa geMaker API calls to AWS CloudTrail. Add", + "D. Use AWS CloudTrail to log Amazon SageMaker API ca lls to Amazon S3. Set up Amazon SNS to" + ], + "correct": "", + "explanation": "To log Amazon SageMaker API calls, the team can use AWS CloudTrail, which is a service that provides a record of actions taken by a user, role, or an AWS service in SageMaker1. CloudTrail captures all API calls for SageMaker, with the exce ption of InvokeEndpoint and InvokeEndpointAsync, as events1. The calls captured include calls from t he SageMaker console and code calls to the SageMaker API operations1. The team can create a tr ail to enable continuous delivery of CloudTrail events to an Amazon S3 bucket, and configure other AWS services to further analyze and act upon the event data collected in CloudTrail logs1. The a uditors can view the CloudTrail log activity report in the CloudTrail console or download the log files fr om the S3 bucket1. To receive a notification when the model is overfit ting, the team can add code to push a custom metric to Amazon CloudWatch, which is a service tha t provides monitoring and observability for AWS resources and applications2. The team can use the M XNet metric API to define and compute the custom metric, such as the validation accuracy or t he validation loss, and use the boto3 CloudWatch client to put the metric data to CloudWatch3 . The team can then create an alarm in CloudWatch with Amazon SNS to receive a notification when the custom metric crosses a threshold that indicates overfitting . For example, the team can set the ala rm to trigger when the validation loss increases fo r a certain number of consecutive periods, which mean s the model is learning the noise in the training data and not generalizing well to the validation da ta.", + "references": "1: Log Amazon SageMaker API Calls with AWS CloudTra il - Amazon SageMaker 2: What Is Amazon CloudWatch? - Amazon CloudWatch 3: Metric API \" Apache MXNet documentation : CloudWatch \" Boto 3 Docs 1.20.21 documentation : Creating Amazon CloudWatch Alarms - Amazon CloudW atch : What is Amazon Simple Notification Service? - Ama zon Simple Notification Service : Overfitting and Underfitting - Machine Learning C rash Course" + }, + { + "question": "A Machine Learning Specialist is implementing a ful l Bayesian network on a dataset that describes public transit in New York City. One of the random variables is discrete, and represents the number of minutes New Yorkers wait for a bus given that the b uses cycle every 10 minutes, with a mean of 3 minutes. Which prior probability distribution should the ML Specialist use for this variable?", + "options": [ + "A. Poisson distribution ,", + "B. Uniform distribution", + "C. Normal distribution", + "D. Binomial distribution" + ], + "correct": "A. Poisson distribution ,", + "explanation": "Explanation/Reference: The prior probability distribution for the discrete random variable that represents the number of minutes New Yorkers wait for a bus is a Poisson dis tribution. A Poisson distribution is suitable for modeling the number of events that occur in a fixed interval of time or space, given a known average rate of occurrence. In this case, the event is wait ing for a bus, the interval is 10 minutes, and the average rate is 3 minutes. The Poisson distribution can capture the variability of the waiting time, which can range from 0 to 10 minutes, with differen t probabilities.", + "references": "1: Poisson Distribution - Amazon SageMaker 2: Poisson Distribution - Wikipedia" + }, + { + "question": "A Data Science team within a large company uses Ama zon SageMaker notebooks to access data stored in Amazon S3 buckets. The IT Security team i s concerned that internet-enabled notebook instances create a security vulnerability where mal icious code running on the instances could compromise data privacy. The company mandates that all instances stay within a secured VPC with no internet access, and data communication traffic must stay within the AWS network. How should the Data Science team configure the note book instance placement to meet these requirements?", + "options": [ + "A. Associate the Amazon SageMaker notebook with a pr ivate subnet in a VPC. Place the Amazon", + "B. Associate the Amazon SageMaker notebook with a pr ivate subnet in a VPC. Use 1AM policies to", + "C. Associate the Amazon SageMaker notebook with a pr ivate subnet in a VPC. Ensure the VPC has S3", + "D. Associate the Amazon SageMaker notebook with a pr ivate subnet in a VPC. Ensure the VPC has a" + ], + "correct": "C. Associate the Amazon SageMaker notebook with a pr ivate subnet in a VPC. Ensure the VPC has S3", + "explanation": "To configure the notebook instance placement to mee t the requirements, the Data Science team should associate the Amazon SageMaker notebook with a private subnet in a VPC. A VPC is a virtual network that is logically isolated from other netwo rks in AWS. A private subnet is a subnet that has no internet gateway attached to it, and therefore c annot communicate with the internet. By placing the notebook instance in a private subnet, the team can ensure that it stays within a secured VPC with no internet access. However, to access data stored in Amazon S3 buckets and other AWS services, the team needs to ensure that the VPC has S3 VPC endpoints and Amazon SageMaker VPC endpoints attached to it. A VPC endpoint is a gateway that enables private conn ections between the VPC and supported AWS services. A VPC endpoint does not require an intern et gateway, a NAT device, or a VPN connection, and ensures that the traffic between the VPC and th e AWS service does not leave the AWS network. By using VPC endpoints, the team can access Amazon S3 and Amazon SageMaker from the notebook instance without compromising data privacy or secur ity.", + "references": ": What Is Amazon VPC? - Amazon Virtual Private Clou d : Subnet Routing - Amazon Virtual Private Cloud : VPC Endpoints - Amazon Virtual Private Cloud" + }, + { + "question": "A Machine Learning Specialist has created a deep le arning neural network model that performs well on the training data but performs poorly on the tes t data. Which of the following methods should the Specialis t consider using to correct this? (Select THREE.) A. Decrease regularization.", + "options": [ + "B. Increase regularization.", + "C. Increase dropout.", + "D. Decrease dropout." + ], + "correct": "", + "explanation": "The problem of poor performance on the test data is a sign of overfitting, which means the model has learned the training data too well and failed t o generalize to new and unseen data. To correct this, the Machine Learning Specialist should consid er using methods that reduce the complexity of the model and increase its ability to generalize. S ome of these methods are: Increase regularization: Regularization is a techni que that adds a penalty term to the loss function o f the model, which reduces the magnitude of the model weights and prevents overfitting. There are different types of regularization, such as L1, L2, and elastic net, that apply different penalties to the weights1. Increase dropout: Dropout is a technique that rando mly drops out some units or connections in the neural network during training, which reduces the c o-dependency of the units and prevents overfitting. Dropout can be applied to different la yers of the network, and the dropout rate can be tuned to control the amount of dropout2. Decrease feature combinations: Feature combinations are the interactions between different input features that can be used to create new features fo r the model. However, too many feature combinations can increase the complexity of the mod el and cause overfitting. Therefore, the Specialist should decrease the number of feature co mbinations and select only the most relevant and informative ones for the model3.", + "references": "1: Regularization for Deep Learning - Amazon SageMa ker 2: Dropout - Amazon SageMaker 3: Feature Engineering - Amazon SageMaker" + }, + { + "question": "A Data Scientist needs to create a serverless inges tion and analytics solution for high-velocity, real time streaming data. The ingestion process must buffer and convert incom ing records from JSON to a query-optimized, columnar format without data loss. The output datas tore must be highly available, and Analysts must be able to run SQL queries against the data and con nect to existing business intelligence dashboards. Which solution should the Data Scientist build to s atisfy the requirements?", + "options": [ + "A. Create a schema in the AWS Glue Data Catalog of t he incoming data format. Use an Amazon", + "B. Write each JSON record to a staging location in A mazon S3. Use the S3 Put event to trigger an", + "C. Write each JSON record to a staging location in A mazon S3. Use the S3 Put event to trigger an AWS" + ], + "correct": "A. Create a schema in the AWS Glue Data Catalog of t he incoming data format. Use an Amazon", + "explanation": "To create a serverless ingestion and analytics solu tion for high-velocity, real-time streaming data, t he Data Scientist should use the following AWS service s: AWS Glue Data Catalog: This is a managed service th at acts as a central metadata repository for data assets across AWS and on-premises data sources. The Data Scientist can use AWS Glue Data Catalog to create a schema of the incoming data format, whi ch defines the structure, format, and data types of the JSON records. The schema can be used by othe r AWS services to understand and process the data1. Amazon Kinesis Data Firehose: This is a fully manag ed service that delivers real-time streaming data to destinations such as Amazon S3, Amazon Redshift, Amazon Elasticsearch Service, and Splunk. The Data Scientist can use Amazon Kinesis Data Firehose to stream the data from the source and transform the data to a query-optimized, columnar f ormat such as Apache Parquet or ORC using the AWS Glue Data Catalog before delivering to Amazon S 3. This enables efficient compression, partitioning, and fast analytics on the data2. Amazon S3: This is an object storage service that o ffers high durability, availability, and scalabilit y. The Data Scientist can use Amazon S3 as the output datastore for the transformed data, which can be organized into buckets and prefixes according to th e desired partitioning scheme. Amazon S3 also integrates with other AWS services such as Amazon A thena, Amazon EMR, and Amazon Redshift Spectrum for analytics3. Amazon Athena: This is a serverless interactive que ry service that allows users to analyze data in Amazon S3 using standard SQL. The Data Scientist ca n use Amazon Athena to run SQL queries against the data in Amazon S3 and connect to existing busin ess intelligence dashboards using the Athena Java Database Connectivity (JDBC) connector. Amazon Athena leverages the AWS Glue Data Catalog to access the schema information and supports forma ts such as Parquet and ORC for fast and costeffecti ve queries4.", + "references": "1: What Is the AWS Glue Data Catalog? - AWS Glue 2: What Is Amazon Kinesis Data Firehose? - Amazon K inesis Data Firehose 3: What Is Amazon S3? - Amazon Simple Storage Servi ce 4: What Is Amazon Athena? - Amazon Athena" + }, + { + "question": "A company is setting up an Amazon SageMaker environ ment. The corporate data security policy does not allow communication over the internet. How can the company enable the Amazon SageMaker ser vice without enabling direct internet access to Amazon SageMaker notebook instances?", + "options": [ + "A. Create a NAT gateway within the corporate VPC.", + "B. Route Amazon SageMaker traffic through an on-prem ises network.", + "C. Create Amazon SageMaker VPC interface endpoints w ithin the corporate VPC.", + "D. Create VPC peering with Amazon VPC hosting Amazon SageMaker." + ], + "correct": "C. Create Amazon SageMaker VPC interface endpoints w ithin the corporate VPC.", + "explanation": "To enable the Amazon SageMaker service without enab ling direct internet access to Amazon SageMaker notebook instances, the company should cr eate Amazon SageMaker VPC interface endpoints within the corporate VPC. A VPC interface endpoint is a gateway that enables private connections between the VPC and supported AWS servi ces without requiring an internet gateway, a NAT device, a VPN connection, or an AWS Direct Conn ect connection. The instances in the VPC do not need to connect to the public internet in order to communicate with the Amazon SageMaker service. The VPC interface endpoint connects the VP C directly to the Amazon SageMaker service using AWS PrivateLink, which ensures that the traff ic between the VPC and the service does not leave the AWS network1.", + "references": "1: Connect to SageMaker Within your VPC - Amazon Sa geMaker" + }, + { + "question": "An office security agency conducted a successful pi lot using 100 cameras installed at key locations within the main office. Images from the cameras wer e uploaded to Amazon S3 and tagged using Amazon Rekognition, and the results were stored in Amazon ES. The agency is now looking to expand the pilot into a full production system using thous ands of video cameras in its office locations globally. The goal is to identify activities perfor med by non-employees in real time. Which solution should the agency consider?", + "options": [ + "A. Use a proxy server at each local office and for e ach camera, and stream the RTSP feed to a unique", + "B. Use a proxy server at each local office and for e ach camera, and stream the RTSP feed to a unique", + "C. Install AWS DeepLens cameras and use the DeepLens _Kinesis_Video module to stream video to", + "D. Install AWS DeepLens cameras and use the DeepLens _Kinesis_Video module to stream video to" + ], + "correct": "A. Use a proxy server at each local office and for e ach camera, and stream the RTSP feed to a unique", + "explanation": "The solution that the agency should consider is to use a proxy server at each local office and for eac h camera, and stream the RTSP feed to a unique Amazon Kinesis Video Streams video stream. On each stream, use Amazon Rekognition Video and create a s tream processor to detect faces from a collection of known employees, and alert when non-e mployees are detected. This solution has the following advantages: It can handle thousands of video cameras in real ti me, as Amazon Kinesis Video Streams can scale elastically to support any number of producers and consumers1. It can leverage the Amazon Rekognition Video API, w hich is designed and optimized for video analysis, and can detect faces in challenging condi tions such as low lighting, occlusions, and differe nt poses2. It can use a stream processor, which is a feature o f Amazon Rekognition Video that allows you to create a persistent application that analyzes strea ming video and stores the results in a Kinesis data stream3. The stream processor can compare the detec ted faces with a collection of known employees, which is a container for persisting face s that you want to search for in the input video stream4. The stream processor can also send notific ations to Amazon Simple Notification Service (Amazon SNS) when non-employees are detected, which can trigger downstream actions such as sending alerts or storing the events in Amazon Elas ticsearch Service (Amazon ES)3.", + "references": "1: What Is Amazon Kinesis Video Streams? - Amazon K inesis Video Streams 2: Detecting and Analyzing Faces - Amazon Rekogniti on 3: Using Amazon Rekognition Video Stream Processor - Amazon Rekognition 4: Working with Stored Faces - Amazon Rekognition" + }, + { + "question": "A financial services company is building a robust s erverless data lake on Amazon S3. The data lake should be flexible and meet the following requireme nts: * Support querying old and new data on Amazon S3 th rough Amazon Athena and Amazon Redshift Spectrum. * Support event-driven ETL pipelines. * Provide a quick and easy way to understand metada ta. Which approach meets trfese requirements?", + "options": [ + "A. Use an AWS Glue crawler to crawl S3 data, an AWS Lambda function to trigger an AWS Glue ETL", + "B. Use an AWS Glue crawler to crawl S3 data, an AWS Lambda function to trigger an AWS Batch job,", + "C. Use an AWS Glue crawler to crawl S3 data, an Amaz on CloudWatch alarm to trigger an AWS Batch", + "D. Use an AWS Glue crawler to crawl S3 data, an Amaz on CloudWatch alarm to trigger an AWS Glue" + ], + "correct": "A. Use an AWS Glue crawler to crawl S3 data, an AWS Lambda function to trigger an AWS Glue ETL", + "explanation": "To build a robust serverless data lake on Amazon S3 that meets the requirements, the financial services company should use the following AWS servi ces: AWS Glue crawler: This is a service that connects t o a data store, progresses through a prioritized li st of classifiers to determine the schema for the data , and then creates metadata tables in the AWS Glue Data Catalog1. The company can use an AWS Glue crawler to crawl the S3 data and infer the schema, format, and partition structure of the data . The crawler can also detect schema changes and update the metadata tables accordingly. This enable s the company to support querying old and new data on Amazon S3 through Amazon Athena and Amazon Redshift Spectrum, which are serverless interactive query services that use the AWS Glue Da ta Catalog as a central location for storing and retrieving table metadata23. AWS Lambda function: This is a service that lets yo u run code without provisioning or managing servers. You pay only for the compute time you cons ume - there is no charge when your code is not running. You can also use AWS Lambda to create even t-driven ETL pipelines, by triggering other AWS services based on events such as object creation or deletion in S3 buckets4. The company can use an AWS Lambda function to trigger an AWS Glue ETL job, which is a serverless way to extract, transform, and load data for analytics. The AWS Glue ETL job c an perform various data processing tasks, such as converting data formats, filtering, aggregating, jo ining, and more. AWS Glue Data Catalog: This is a managed service th at acts as a central metadata repository for data assets across AWS and on-premises data sources. The AWS Glue Data Catalog provides a uniform repository where disparate systems can store and fi nd metadata to keep track of data in data silos, and use that metadata to query and transform the da ta. The company can use the AWS Glue Data Catalog to search and discover metadata, such as ta ble definitions, schemas, and partitions. The AWS Glue Data Catalog also integrates with Amazon Athen a, Amazon Redshift Spectrum, Amazon EMR, and AWS Glue ETL jobs, providing a consistent view of the data across different query and analysis services. Reference: 1: What Is a Crawler? - AWS Glue 2: What Is Amazon Athena? - Amazon Athena 3: Amazon Redshift Spectrum - Amazon Redshift 4: What is AWS Lambda? - AWS Lambda : AWS Glue ETL Jobs - AWS Glue : What Is the AWS Glue Data Catalog? - AWS Glue", + "references": "" + }, + { + "question": "A company's Machine Learning Specialist needs to im prove the training speed of a time-series forecasting model using TensorFlow. The training is currently implemented on a single-GPU machine and takes approximately 23 hours to complete. The t raining needs to be run daily. The model accuracy js acceptable, but the company a nticipates a continuous increase in the size of the training data and a need to update the model on an hourly, rather than a daily, basis. The company also wants to minimize coding effort and in frastructure changes What should the Machine Learning Specialist do to t he training solution to allow it to scale for futur e demand?", + "options": [ + "A. Do not change the TensorFlow code. Change the mac hine to one with a more powerful GPU to", + "B. Change the TensorFlow code to implement a Horovod distributed framework supported by", + "C. Switch to using a built-in AWS SageMaker DeepAR m odel. Parallelize the training to as many", + "D. Move the training to Amazon EMR and distribute th e workload to as many machines as needed to" + ], + "correct": "B. Change the TensorFlow code to implement a Horovod distributed framework supported by", + "explanation": "To improve the training speed of a time-series fore casting model using TensorFlow, the Machine Learning Specialist should change the TensorFlow co de to implement a Horovod distributed framework supported by Amazon SageMaker. Horovod is a free and open-source software framework for distributed deep learning training us ing TensorFlow, Keras, PyTorch, and Apache MXNet1. Horovod can scale up to hundreds of GPUs wi th upwards of 90% scaling efficiency2. Horovod is easy to use, as it requires only a few lines of Python code to modify an existing training script2. Horovod is also portable , as it runs the same for TensorFlow, Keras, PyTorc h, and MXNet; on premise, in the cloud, and on Apache Spark2. Amazon SageMaker is a fully managed service that pr ovides every developer and data scientist with the ability to build, train, and deploy machine lea rning models quickly3. Amazon SageMaker supports Horovod as a built-in distributed training framework, which means that the Machine Learning Specialist does not need to install or con figure Horovod separately4. Amazon SageMaker also provides a number of features and tools to sim plify and optimize the distributed training process, such as automatic scaling, debugging, prof iling, and monitoring4. By using Amazon SageMaker, the Machine Learning Specialist can para llelize the training to as many machines as needed to achieve the business goals, while minimiz ing coding effort and infrastructure changes.", + "references": "1: Horovod (machine learning) - Wikipedia 2: Home - Horovod 3: Amazon SageMaker \" Machine Learning Service \" AW S 4: Use Horovod with Amazon SageMaker - Amazon SageM aker" + }, + { + "question": "A Machine Learning Specialist is required to build a supervised image-recognition model to identify a cat. The ML Specialist performs some tests and reco rds the following results for a neural networkbasedimage classifier: Total number of images available = 1,000 Test set i mages = 100 (constant test set) The ML Specialist notices that, in over 75% of the misclassified images, the cats were held upside down by their owners. Which techniques can be used by the ML Specialist t o improve this specific test error?", + "options": [ + "A. Increase the training data by adding variation in rotation for training images.", + "B. Increase the number of epochs for model training.", + "C. Increase the number of layers for the neural netw ork.", + "D. Increase the dropout rate for the second-to-last layer." + ], + "correct": "A. Increase the training data by adding variation in rotation for training images.", + "explanation": "To improve the test error for the image classifier, the Machine Learning Specialist should use the technique of increasing the training data by adding variation in rotation for training images. This technique is called data augmentation, which is a w ay of artificially expanding the size and diversity of the training dataset by applying various transfo rmations to the original images, such as rotation, flipping, cropping, scaling, etc. Data augmentation can help the model learn more robust features that are invariant to the orientation, position, an d size of the objects in the images. This can impro ve the generalization ability of the model and reduce the test error, especially for cases where the images are not well-aligned or have different persp ectives1.", + "references": "1: Image Augmentation - Amazon SageMaker" + }, + { + "question": "A Data Scientist is developing a machine learning m odel to classify whether a financial transaction is fraudulent. The labeled data available for training consists of 100,000 non-fraudulent observations and 1,000 fraudulent observations. The Data Scientist applies the XGBoost algorithm to the data, resulting in the following confusion matrix when the trained model is applied to a previ ously unseen validation dataset. The accuracy of the model is 99.1%, but the Data Scientist has been asked to reduce the number of false negatives. Which combination of steps should the Data Scientis t take to reduce the number of false positive predictions by the model? (Select TWO.)", + "options": [ + "A. Change the XGBoost eval_metric parameter to optim ize based on rmse instead of error.", + "B. Increase the XGBoost scale_pos_weight parameter t o adjust the balance of positive and negative", + "C. Increase the XGBoost max_depth parameter because the model is currently underfitting the data.", + "D. Change the XGBoost evaljnetric parameter to optim ize based on AUC instead of error." + ], + "correct": "", + "explanation": "The XGBoost algorithm is a popular machine learning technique for classification problems. It is based on the idea of boosting, which is to combine many weak learners (decision trees) into a strong learner (ensemble model). The XGBoost algorithm can handle imbalanced data by using the scale_pos_weight parameter, which controls the balance of positive and negative weigh ts in the objective function. A typical value to consider is the ratio of negative cases to positive cases in the data. By increasing this parameter, t he algorithm will pay more attention to the minority c lass (positive) and reduce the number of false negatives. The XGBoost algorithm can also use different evalua tion metrics to optimize the model performance. The default metric is error, which is the misclassi fication rate. However, this metric can be misleadi ng for imbalanced data, as it does not account for the different costs of false positives and false negatives. A better metric to use is AUC, which is the area under the receiver operating characteristi c (ROC) curve. The ROC curve plots the true positive rate against the false positive rate for different threshold values. The AUC measures how well the mod el can distinguish between the two classes, regardless of the threshold. By changing the eval_m etric parameter to AUC, the algorithm will try to maximize the AUC score and reduce the number of fal se negatives. Therefore, the combination of steps that should be taken to reduce the number of false negatives are to increase the scale_pos_weight parameter and chan ge the eval_metric parameter to AUC.", + "references": "XGBoost Parameters XGBoost for Imbalanced Classification" + }, + { + "question": "A Machine Learning Specialist is assigned a TensorF low project using Amazon SageMaker for training, and needs to continue working for an extended perio d with no Wi-Fi access. Which approach should the Specialist use to continu e working?", + "options": [ + "A. Install Python 3 and boto3 on their laptop and co ntinue the code development using that", + "B. Download the TensorFlow Docker container used in Amazon SageMaker from GitHub to their local", + "C. Download TensorFlow from tensorflow.org to emulat e the TensorFlow kernel in the SageMaker", + "D. Download the SageMaker notebook to their local en vironment then install Jupyter Notebooks on" + ], + "correct": "B. Download the TensorFlow Docker container used in Amazon SageMaker from GitHub to their local", + "explanation": "Amazon SageMaker is a fully managed service that en ables developers and data scientists to quickly and easily build, train, and deploy machine learnin g models at any scale. SageMaker provides a variety of tools and frameworks to support the enti re machine learning workflow, from data preparation to model deployment. One of the tools that SageMaker offers is the Amazo n SageMaker Python SDK, which is a high-level library that simplifies the interaction with SageMa ker APIs and services. The SageMaker Python SDK allows you to write code in Python and use popular frameworks such as TensorFlow, PyTorch, MXNet, and more. You can use the SageMaker Python SDK to c reate and manage SageMaker resources such as notebook instances, training jobs, endpoints, an d feature store. If you need to continue working on a TensorFlow pro ject using SageMaker for training without Wi-Fi access, the best approach is to download the Tensor Flow Docker container used in SageMaker from GitHub to your local environment, and use the SageM aker Python SDK to test the code. This way, you can ensure that your code is compatible with the Sa geMaker environment and avoid any potential issues when you upload your code to SageMaker and s tart the training job. You can also use the same code to deploy your model to a SageMaker endpoint w hen you have Wi-Fi access again. To download the TensorFlow Docker container used in SageMaker, you can visit the SageMaker Docker GitHub repository and follow the instruction s to build the image locally. You can also use the SageMaker Studio Image Build CLI to automate th e process of building and pushing the Docker image to Amazon Elastic Container Registry (Amazon ECR). To use the SageMaker Python SDK to test the code, you can install the SDK on your local mac hine by following the installation guide. You can also refer to the TensorFlow documentation for more details on how to use the SageMaker Python SDK with TensorFlow.", + "references": "SageMaker Docker GitHub repository SageMaker Studio Image Build CLI SageMaker Python SDK installation guide SageMaker Python SDK TensorFlow documentation" + }, + { + "question": "A Data Scientist wants to gain real-time insights i nto a data stream of GZIP files. Which solution would allow the use of SQL to query the stream with the LEAST latency?", + "options": [ + "A. Amazon Kinesis Data Analytics with an AWS Lambda function to transform the data.", + "B. AWS Glue with a custom ETL script to transform th e data.", + "C. An Amazon Kinesis Client Library to transform the data and save it to an Amazon ES cluster.", + "D. Amazon Kinesis Data Firehose to transform the dat a and put it into an Amazon S3 bucket." + ], + "correct": "A. Amazon Kinesis Data Analytics with an AWS Lambda function to transform the data.", + "explanation": "Amazon Kinesis Data Analytics is a service that ena bles you to analyze streaming data in real time using SQL or Apache Flink applications. You can use Kinesis Data Analytics to process and gain insights from data streams such as web logs, clicks treams, IoT data, and more. To use SQL to query a data stream of GZIP files, yo u need to first transform the data into a format th at Kinesis Data Analytics can understand, such as JSON , CSV, or Apache Parquet. You can use an AWS Lambda function to perform this transformation and send the output to a Kinesis data stream that is connected to your Kinesis Data Analytics applicatio n. This way, you can use SQL to query the stream with the least latency, as Lambda functions are tri ggered in near real time by the incoming data and Kinesis Data Analytics can process the data as soon as it arrives. The other options are not optimal for this scenario , as they introduce more latency or complexity. AWS Glue is a serverless data integration service t hat can perform ETL (extract, transform, and load) tasks on data sources, but it is not designed for r eal-time streaming data analysis. An Amazon Kinesis Client Library is a Java library that enables you t o build custom applications that process data from Kinesis data streams, but it requires more coding a nd configuration than using a Lambda function. Amazon Kinesis Data Firehose is a service that can deliver streaming data to destinations such as Amazon S3, Amazon Redshift, Amazon OpenSearch Servi ce, and Splunk, but it does not support SQL queries on the data.", + "references": "What Is Amazon Kinesis Data Analytics for SQL Appli cations? Using AWS Lambda with Amazon Kinesis Data Streams Using AWS Lambda with Amazon Kinesis Data Firehose" + }, + { + "question": "A Machine Learning Specialist must build out a proc ess to query a dataset on Amazon S3 using Amazon Athena The dataset contains more than 800.00 0 records stored as plaintext CSV files Each record contains 200 columns and is approximately 1 5 MB in size Most queries will span 5 to 10 columns only How should the Machine Learning Specialist transfor m the dataset to minimize query runtime? A. Convert the records to Apache Parquet format", + "options": [ + "B. Convert the records to JSON format", + "C. Convert the records to GZIP CSV format", + "D. Convert the records to XML format" + ], + "correct": "", + "explanation": "Amazon Athena is an interactive query service that allows you to analyze data stored in Amazon S3 using standard SQL. Athena is serverless, so you on ly pay for the queries that you run and there is no infrastructure to manage. To optimize the query performance of Athena, one of the best practices is to convert the data into a columnar format, such as Apache Parquet or Apache O RC. Columnar formats store data by columns rather than by rows, which allows Athena to scan on ly the columns that are relevant to the query, reducing the amount of data read and improving the query speed. Columnar formats also support compression and encoding schemes that can reduce th e storage space and the data scanned per query, further enhancing the performance and reduci ng the cost. In contrast, plaintext CSV files store data by rows , which means that Athena has to scan the entire row even if only a few columns are needed for the q uery. This increases the amount of data read and the query latency. Moreover, plaintext CSV files do not support compression or encoding, which means that they take up more storage space and incu r higher query costs. Therefore, the Machine Learning Specialist should t ransform the dataset to Apache Parquet format to minimize query runtime.", + "references": "https://www.cloudforecast.io/blog/using- parquet-on-athena-to-save-money-on-aws/" + }, + { + "question": "A Machine Learning Specialist is developing a daily ETL workflow containing multiple ETL jobs The workflow consists of the following processes * Start the workflow as soon as data is uploaded to Amazon S3 * When all the datasets are available in Amazon S3, start an ETL job to join the uploaded datasets with multiple terabyte-sized datasets already store d in Amazon S3 * Store the results of joining datasets in Amazon S 3 * If one of the jobs fails, send a notification to the Administrator Which configuration will meet these requirements?", + "options": [ + "A. Use AWS Lambda to trigger an AWS Step Functions w orkflow to wait for dataset uploads to", + "B. Develop the ETL workflow using AWS Lambda to star t an Amazon SageMaker notebook instance", + "C. Develop the ETL workflow using AWS Batch to trigg er the start of ETL jobs when data is uploaded", + "D. Use AWS Lambda to chain other Lambda functions to read and join the datasets in Amazon S3 as" + ], + "correct": "", + "explanation": "To develop a daily ETL workflow containing multiple ETL jobs that can start as soon as data is uploaded to Amazon S3, the best configuration is to use AWS Lambda to trigger an AWS Step Functions workflow to wait for dataset uploads to c omplete in Amazon S3. Use AWS Glue to join the datasets. Use an Amazon CloudWatch alarm to send an SNS notification to the Administrator in the case of a failure. AWS Lambda is a serverless compute service that let s you run code without provisioning or managing servers. You can use Lambda to create func tions that respond to events such as data uploads to Amazon S3. You can also use Lambda to in voke other AWS services such as AWS Step Functions and AWS Glue. AWS Step Functions is a service that lets you coord inate multiple AWS services into serverless workflows. You can use Step Functions to create a s tate machine that defines the sequence and logic of your ETL workflow. You can also use Step Functio ns to handle errors and retries, and to monitor the execution status of your workflow. AWS Glue is a serverless data integration service t hat makes it easy to discover, prepare, and combine data for analytics. You can use Glue to cre ate and run ETL jobs that can join data from multiple sources in Amazon S3. You can also use Glu e to catalog your data and make it searchable and queryable. Amazon CloudWatch is a service that monitors your A WS resources and applications. You can use CloudWatch to create alarms that trigger actions wh en a metric or a log event meets a specified threshold. You can also use CloudWatch to send noti fications to Amazon Simple Notification Service (SNS) topics, which can then deliver the notificati ons to subscribers such as email addresses or phone numbers. Therefore, by using these services together, you ca n achieve the following benefits: You can start the ETL workflow as soon as data is u ploaded to Amazon S3 by using Lambda functions to trigger Step Functions workflows. You can wait for all the datasets to be available i n Amazon S3 by using Step Functions to poll the S3 buckets and check the data completeness. You can join the datasets with terabyte-sized datas ets in Amazon S3 by using Glue ETL jobs that can scale and parallelize the data processing. You can store the results of joining datasets in Am azon S3 by using Glue ETL jobs to write the output to S3 buckets. You can send a notification to the Administrator if one of the jobs fails by using CloudWatch alarms t o monitor the Step Functions or Glue metrics and send SNS notifications in case of a failure.", + "references": "" + }, + { + "question": "An agency collects census information within a coun try to determine healthcare and social program needs by province and city. The census form collect s responses for approximately 500 questions from each citizen Which combination of algorithms would provide the a ppropriate insights? (Select TWO )", + "options": [ + "A. The factorization machines (FM) algorithm", + "B. The Latent Dirichlet Allocation (LDA) algorithm", + "C. The principal component analysis (PCA) algorithm", + "D. The k-means algorithm" + ], + "correct": "", + "explanation": "The agency wants to analyze the census data for pop ulation segmentation, which is a type of unsupervised learning problem that aims to group si milar data points together based on their attributes. The agency can use a combination of alg orithms that can perform dimensionality reduction and clustering on the data to achieve thi s goal. Dimensionality reduction is a technique that reduce s the number of features or variables in a dataset while preserving the essential information and rela tionships. Dimensionality reduction can help improve the efficiency and performance of clusterin g algorithms, as well as facilitate data visualization and interpretation. One of the most c ommon algorithms for dimensionality reduction is principal component analysis (PCA), which transform s the original features into a new set of orthogonal features called principal components tha t capture the maximum variance in the data. PCA can help reduce the noise and redundancy in the data and reveal the underlying structure and patterns. Clustering is a technique that partitions the data into groups or clusters based on their similarity o r distance. Clustering can help discover the natural segments or categories in the data and understand their characteristics and differences. One of the m ost popular algorithms for clustering is k-means, which assigns each data point to one of k clusters based on the nearest mean or centroid. K-means can handle large and high-dimensional datasets and produce compact and spherical clusters. Therefore, the combination of algorithms that would provide the appropriate insights for population segmentation are PCA and k-means. The agency can us e PCA to reduce the dimensionality of the census data from 500 features to a smaller number o f principal components that capture most of the variation in the data. Then, the agency can use k-m eans to cluster the data based on the principal components and identify the segments of the populat ion that share similar characteristics.", + "references": "Amazon SageMaker Principal Component Analysis (PCA) Amazon SageMaker K-Means Algorithm" + }, + { + "question": "A large consumer goods manufacturer has the followi ng products on sale 34 different toothpaste variants 48 different toothbrush variants 43 different mouthwash variants The entire sales history of all these products is a vailable in Amazon S3 Currently, the company is using custom-built autoregressive integrated moving average (ARIMA) models to forecast demand for these products The company wants to predict the demand for a new product that will soon be launched Which solution should a Machine Learning Specialist apply?", + "options": [ + "A. Train a custom ARIMA model to forecast demand for the new product.", + "B. Train an Amazon SageMaker DeepAR algorithm to for ecast demand for the new product", + "C. Train an Amazon SageMaker k-means clustering algo rithm to forecast demand for the new", + "D. Train a custom XGBoost model to forecast demand f or the new product" + ], + "correct": "B. Train an Amazon SageMaker DeepAR algorithm to for ecast demand for the new product", + "explanation": "The company wants to predict the demand for a new p roduct that will soon be launched, based on the sales history of similar products. This is a ti me series forecasting problem, which requires a machine learning algorithm that can learn from hist orical data and generate future predictions. One of the most suitable solutions for this problem is to use the Amazon SageMaker DeepAR algorithm, which is a supervised learning algorithm for forecasting scalar time series using recurrent neural networks (RNN). DeepAR can handle multiple r elated time series, such as the sales of different products, and learn a global model that c aptures the common patterns and trends across the time series. DeepAR can also generate probabili stic forecasts that provide confidence intervals and quantify the uncertainty of the predictions. DeepAR can outperform traditional forecasting metho ds, such as ARIMA, especially when the dataset contains hundreds or thousands of related t ime series. DeepAR can also use the trained model to forecast the demand for new products that are similar to the ones it has been trained on, by using the categorical features that encode the p roduct attributes. For example, the company can use the product type, brand, flavor, size, and pric e as categorical features to group the products and learn the typical behavior for each group. Therefore, the Machine Learning Specialist should a pply the Amazon SageMaker DeepAR algorithm to forecast the demand for the new product, by usin g the sales history of the existing products as the training dataset, and the product attributes as the categorical features.", + "references": "DeepAR Forecasting Algorithm - Amazon SageMaker Now available in Amazon SageMaker: DeepAR algorithm for more accurate time series forecasting" + }, + { + "question": "A Data Scientist needs to migrate an existing on-pr emises ETL process to the cloud The current process runs at regular time intervals and uses PyS park to combine and format multiple large data sources into a single consolidated output for downs tream processing The Data Scientist has been given the following req uirements for the cloud solution * Combine multiple data sources * Reuse existing PySpark logic * Run the solution on the existing schedule * Minimize the number of servers that will need to be managed Which architecture should the Data Scientist use to build this solution?", + "options": [ + "A. Write the raw data to Amazon S3 Schedule an AWS L ambda function to submit a Spark step to a", + "B. Write the raw data to Amazon S3 Create an AWS Glu e ETL job to perform the ETL processing", + "C. Write the raw data to Amazon S3 Schedule an AWS L ambda function to run on the existing", + "D. Use Amazon Kinesis Data Analytics to stream the i nput data and perform realtime SQL queries" + ], + "correct": "B. Write the raw data to Amazon S3 Create an AWS Glu e ETL job to perform the ETL processing", + "explanation": "The Data Scientist needs to migrate an existing on- premises ETL process to the cloud, using a solution that can combine multiple data sources, reuse exist ing PySpark logic, run on the existing schedule, and minimize the number of servers that need to be managed. The best architecture for this scenario is to use AWS Glue, which is a serverless data inte gration service that can create and run ETL jobs on AWS. AWS Glue can perform the following tasks to meet th e requirements: Combine multiple data sources: AWS Glue can access data from various sources, such as Amazon S3, Amazon RDS, Amazon Redshift, Amazon DynamoDB, and m ore. AWS Glue can also crawl the data sources and discover their schemas, formats, and pa rtitions, and store them in the AWS Glue Data Catalog, which is a centralized metadata repository for all the data assets. Reuse existing PySpark logic: AWS Glue supports wri ting ETL scripts in Python or Scala, using Apache Spark as the underlying execution engine. AWS Glue provides a library of built-in transformations and connectors that can simplify the ETL code. The Data Scientist can write the ETL job in PySpark and leverage the existing logic to perform the data processing. Run the solution on the existing schedule: AWS Glue can create triggers that can start ETL jobs based on a schedule, an event, or a condition. The Data S cientist can create a new AWS Glue trigger to run the ETL job based on the existing schedule, using a cron expression or a relative time interval. Minimize the number of servers that need to be mana ged: AWS Glue is a serverless service, which means that it automatically provisions, configures, scales, and manages the compute resources required to run the ETL jobs. The Data Scientist do es not need to worry about setting up, maintaining, or monitoring any servers or clusters for the ETL process. Therefore, the Data Scientist should use the follow ing architecture to build the cloud solution: Write the raw data to Amazon S3: The Data Scientist can use any method to upload the raw data from the on-premises sources to Amazon S3, such as AWS DataSync, AWS Storage Gateway, AWS Snowball, or AWS Direct Connect. Amazon S3 is a dur able, scalable, and secure object storage service that can store any amount and type of data. Create an AWS Glue ETL job to perform the ETL proce ssing against the input data: The Data Scientist can use the AWS Glue console, AWS Glue API, AWS SDK , or AWS CLI to create and configure an AWS Glue ETL job. The Data Scientist can specify the in put and output data sources, the IAM role, the security configuration, the job parameters, and the PySpark script location. The Data Scientist can also use the AWS Glue Studio, which is a graphical interface that can help design, run, and monitor ETL jobs visually. Write the ETL job in PySpark to leverage the existi ng logic: The Data Scientist can use a code editor of their choice to write the ETL script in PySpark, us ing the existing logic to transform the data. The D ata Scientist can also use the AWS Glue script editor, which is an integrated development environment (IDE) that can help write, debug, and test the ETL code. The Data Scientist can store the ETL script i n Amazon S3 or GitHub, and reference it in the AWS Gl ue ETL job configuration. Create a new AWS Glue trigger to trigger the ETL jo b based on the existing schedule: The Data Scientist can use the AWS Glue console, AWS Glue AP I, AWS SDK, or AWS CLI to create and configure an AWS Glue trigger. The Data Scientist can specify the name, type, and schedule of the trigger, and associate it with the AWS Glue ETL job. The trigger will start the ETL job according to the defined schedule. Configure the output target of the ETL job to write to a oeprocessed location in Amazon S3 that is accessible for downstream use: The Data Scientist c an specify the output location of the ETL job in the PySpark script, using the AWS Glue DynamicFrame or Spark DataFrame APIs. The Data Scientist can write the output data to a oeprocessed location in Amazon S3, using a format such as Parquet, ORC, JSON, or CSV, that is suitable for downstream processing.", + "references": "What Is AWS Glue? AWS Glue Components AWS Glue Studio AWS Glue Triggers" + }, + { + "question": "A large company has developed a B1 application that generates reports and dashboards using data collected from various operational metrics The comp any wants to provide executives with an enhanced experience so they can use natural languag e to get data from the reports The company wants the executives to be able ask questions using written and spoken interlaces Which combination of services can be used to build this conversational interface? (Select THREE)", + "options": [ + "A. Alexa for Business", + "B. Amazon Connect", + "C. Amazon Lex", + "D. Amazon Poly" + ], + "correct": "", + "explanation": "Explanation/Reference: To build a conversational interface that can use na tural language to get data from the reports, the company can use a combination of services that can handle both written and spoken inputs, understand the users intent and query, and extract the relevant information from the reports. The services that can be used for this purpose are: Amazon Lex: A service for building conversational i nterfaces into any application using voice and text . Amazon Lex can create chatbots that can interact wi th users using natural language, and integrate with other AWS services such as Amazon Connect, Ama zon Comprehend, and Amazon Transcribe. Amazon Lex can also use lambda functions to impleme nt the business logic and fulfill the users requests. Amazon Comprehend: A service for natural language p rocessing and text analytics. Amazon Comprehend can analyze text and speech inputs and e xtract insights such as entities, key phrases, sentiment, syntax, and topics. Amazon Comprehend ca n also use custom classifiers and entity recognizers to identify specific terms and concepts that are relevant to the domain of the reports. Amazon Transcribe: A service for speech-to-text con version. Amazon Transcribe can transcribe audio inputs into text outputs, and add punctuation and f ormatting. Amazon Transcribe can also use custom vocabularies and language models to improve the accuracy and quality of the transcription for the specific domain of the reports. Therefore, the company can use the following archit ecture to build the conversational interface: Use Amazon Lex to create a chatbot that can accept both written and spoken inputs from the executives. The chatbot can use intents, utterances , and slots to capture the users query and parameters, such as the report name, date, metric, or filter. Use Amazon Transcribe to convert the spoken inputs into text outputs, and pass them to Amazon Lex. Amazon Transcribe can use a custom vocabulary and l anguage model to recognize the terms and concepts related to the reports. Use Amazon Comprehend to analyze the text inputs an d outputs, and extract the relevant information from the reports. Amazon Comprehend can use a custom classifier and entity recognizer to identify the report name, date, metric, or filte r from the users query, and the corresponding data from the reports. Use a lambda function to implement the business log ic and fulfillment of the users query, such as retrieving the data from the reports, performing ca lculations or aggregations, and formatting the response. The lambda function can also handle error s and validations, and provide feedback to the user. Use Amazon Lex to return the response to the user, either in text or speech format, depending on the users preference.", + "references": "What Is Amazon Lex? What Is Amazon Comprehend? What Is Amazon Transcribe?" + }, + { + "question": "A Machine Learning Specialist is applying a linear least squares regression model to a dataset with 1 000 records and 50 features Prior to training, the ML Specialist notices that two features are perfect ly linearly dependent Why could this be an issue for the linear least squ ares regression model?", + "options": [ + "A. It could cause the backpropagation algorithm to f ail during training", + "B. It could create a singular matrix during optimiza tion which fails to define a unique solution", + "C. It could modify the loss function during optimiza tion causing it to fail during training", + "D. It could introduce non-linear dependencies within the data which could invalidate the linear" + ], + "correct": "B. It could create a singular matrix during optimiza tion which fails to define a unique solution", + "explanation": "Linear least squares regression is a method of fitt ing a linear model to a set of data by minimizing t he sum of squared errors between the observed and pred icted values. The solution of the linear least squares problem can be obtained by solving the norm al equations, which are given by ATAx=ATb, where A is the matrix of explanatory variables, b i s the vector of response variables, and x is the vector of unknown coefficients. However, if the matrix A has two features that are perfectly linearly dependent, then the matrix ATA will be singular, meaning that it does not have a u nique inverse. This implies that the normal equations do not have a unique solution, and the li near least squares problem is ill-posed. In other words, there are infinitely many values of x that c an satisfy the normal equations, and the linear model is not identifiable. This can be an issue for the linear least squares r egression model, as it can lead to instability, inconsistency, and poor generalization of the model . It can also cause numerical difficulties when trying to solve the normal equations using computat ional methods, such as matrix inversion or decomposition. Therefore, it is advisable to avoid or remove the linearly dependent features from the matrix A before applying the linear least squar es regression model.", + "references": "Linear least squares (mathematics) Linear Regression in Matrix Form Singular Matrix Problem" + }, + { + "question": "A Machine Learning Specialist uploads a dataset to an Amazon S3 bucket protected with server-side encryption using AWS KMS. How should the ML Specialist define the Amazon Sage Maker notebook instance so it can read the same dataset from Amazon S3?", + "options": [ + "A. Define security group(s) to allow all HTTP inboun d/outbound traffic and assign those security", + "B. \u00d0\u00a1onfigure the Amazon SageMaker notebook instance to have access to the VPC. Grant permission", + "C. Assign an IAM role to the Amazon SageMaker notebo ok with S3 read access to the dataset. Grant", + "D. Assign the same KMS key used to encrypt data in A mazon S3 to the Amazon SageMaker notebook" + ], + "correct": "C. Assign an IAM role to the Amazon SageMaker notebo ok with S3 read access to the dataset. Grant", + "explanation": "To read data from an Amazon S3 bucket that is prote cted with server-side encryption using AWS KMS, the Amazon SageMaker notebook instance needs t o have an IAM role that has permission to access the S3 bucket and the KMS key. The IAM role is an identity that defines the permissions for the notebook instance to interact with other AWS se rvices. The IAM role can be assigned to the notebook instance when it is created or updated lat er. The KMS key policy is a document that specifies who can use and manage the KMS key. The KMS key policy can grant permission to the IAM role of the notebook instance to decrypt the data in the S3 bucket. The KMS key policy can also grant permissio n to other principals, such as AWS accounts, IAM users, or IAM roles, to use the KMS key for encrypt ion and decryption operations. Therefore, the Machine Learning Specialist should a ssign an IAM role to the Amazon SageMaker notebook with S3 read access to the dataset. Grant permission in the KMS key policy to that role. This way, the notebook instance can use the IAM rol e credentials to access the S3 bucket and the KMS key, and read the encrypted data from the S3 bu cket.", + "references": "Create an IAM Role to Grant Permissions to Your Not ebook Instance Using Key Policies in AWS KMS" + }, + { + "question": "A Data Scientist is building a model to predict cus tomer churn using a dataset of 100 continuous numerical features. The Marketing team has not provided any i nsight about which features are relevant for churn prediction. The Marketing team wants to interpret t he model and see the direct impact of relevant features on the model outcome. While training a logistic regres sion model, the Data Scientist observes that there is a wide gap between the training and validation set accurac y. Which methods can the Data Scientist use to improve the model performance and satisfy the Marketing teams needs? (Choose two.)", + "options": [ + "A. Add L1 regularization to the classifier", + "B. Add features to the dataset", + "C. Perform recursive feature elimination", + "D. Perform t-distributed stochastic neighbor embeddi ng (t-SNE)" + ], + "correct": "", + "explanation": "The Data Scientist is building a model to predict c ustomer churn using a dataset of 100 continuous numerical features. The Marketing team wants to int erpret the model and see the direct impact of relevant features on the model outcome. However, th e Data Scientist observes that there is a wide gap between the training and validation set accurac y, which indicates that the model is overfitting the data and generalizing poorly to new data. To improve the model performance and satisfy the Ma rketing teams needs, the Data Scientist can use the following methods: Add L1 regularization to the classifier: L1 regular ization is a technique that adds a penalty term to the loss function of the logistic regression model, pro portional to the sum of the absolute values of the coefficients. L1 regularization can help reduce ove rfitting by shrinking the coefficients of the less important features to zero, effectively performing feature selection. This can simplify the model and make it more interpretable, as well as improve the validation accuracy. Perform recursive feature elimination: Recursive fe ature elimination (RFE) is a feature selection technique that involves training a model on a subse t of the features, and then iteratively removing the least important features one by one until the d esired number of features is reached. The idea behind RFE is to determine the contribution of each feature to the model by measuring how well the model performs when that feature is removed. The fe atures that are most important to the model will have the greatest impact on performance when t hey are removed. RFE can help improve the model performance by eliminating the irrelevant or redundant features that may cause noise or multicollinearity in the data. RFE can also help th e Marketing team understand the direct impact of the relevant features on the model outcome, as the remaining features will have the highest weights in the model.", + "references": "Regularization for Logistic Regression Recursive Feature Elimination" + }, + { + "question": "An aircraft engine manufacturing company is measuri ng 200 performance metrics in a time-series. Engineers want to detect critical manufacturing defects in ne ar-real time during testing. All of the data needs to be stored for offline analysis. What approach would be the MOST effective to perfor m near-real time defect detection?", + "options": [ + "A. Use AWS IoT Analytics for ingestion, storage, and further analysis. Use Jupyter notebooks from", + "B. Use Amazon S3 for ingestion, storage, and further analysis. Use an Amazon EMR cluster to carry", + "C. Use Amazon S3 for ingestion, storage, and further analysis. Use the Amazon SageMaker Random", + "D. Use Amazon Kinesis Data Firehose for ingestion an d Amazon Kinesis Data Analytics Random Cut" + ], + "correct": "D. Use Amazon Kinesis Data Firehose for ingestion an d Amazon Kinesis Data Analytics Random Cut", + "explanation": "The company wants to perform near-real time defect detection on a time-series of 200 performance metrics, and store all the data for offline analysi s. The best approach for this scenario is to use Amazon Kinesis Data Firehose for ingestion and Amaz on Kinesis Data Analytics Random Cut Forest (RCF) to perform anomaly detection. Use Kinesis Dat a Firehose to store data in Amazon S3 for further analysis. Amazon Kinesis Data Firehose is a service that can capture, transform, and deliver streaming data to destinations such as Amazon S3, Amazon Redshift, Am azon OpenSearch Service, and Splunk. Kinesis Data Firehose can handle any amount and frequency o f data, and automatically scale to match the throughput. Kinesis Data Firehose can also compress , encrypt, and batch the data before delivering it to the destination, reducing the storage cost and e nhancing the security. Amazon Kinesis Data Analytics is a service that can analyze streaming data in real time using SQL or Apache Flink applications. Kinesis Data Analytics c an use built-in functions and algorithms to perform various analytics tasks, such as aggregations, join s, filters, windows, and anomaly detection. One of the built-in algorithms that Kinesis Data Analytics supports is Random Cut Forest (RCF), which is a supervised learning algorithm for forecasting scala r time series using recurrent neural networks. RCF can detect anomalies in streaming data by assigning an anomaly score to each data point, based on how distant it is from the rest of the data. RCF ca n handle multiple related time series, such as the performance metrics of the aircraft engine, and lea rn a global model that captures the common patterns and trends across the time series. Therefore, the company can use the following archit ecture to build the near-real time defect detection solution: Use Amazon Kinesis Data Firehose for ingestion: The company can use Kinesis Data Firehose to capture the streaming data from the aircraft engine testing, and deliver it to two destinations: Amazon S3 and Amazon Kinesis Data Analytics. The co mpany can configure the Kinesis Data Firehose delivery stream to specify the source, the buffer s ize and interval, the compression and encryption options, the error handling and retry logic, and th e destination details. Use Amazon Kinesis Data Analytics Random Cut Forest (RCF) to perform anomaly detection: The company can use Kinesis Data Analytics to create a SQL application that can read the streaming data from the Kinesis Data Firehose delivery stream, and apply the RCF algorithm to detect anomalies. The company can use the RANDOM_CUT_FOREST or RANDOM _CUT_FOREST_WITH_EXPLANATION functions to compute the anomaly scores and attribu tions for each data point, and use the WHERE clause to filter out the normal data points. The co mpany can also use the CURSOR function to specify the input stream, and the PUMP function to write th e output stream to another destination, such as Amazon Kinesis Data Streams or AWS Lambda. Use Kinesis Data Firehose to store data in Amazon S 3 for further analysis: The company can use Kinesis Data Firehose to store the raw and processe d data in Amazon S3 for offline analysis. The company can use the S3 destination of the Kinesis D ata Firehose delivery stream to store the raw data, and use another Kinesis Data Firehose deliver y stream to store the output of the Kinesis Data Analytics application. The company can also use AWS Glue or Amazon Athena to catalog, query, and analyze the data in Amazon S3.", + "references": "What Is Amazon Kinesis Data Firehose? What Is Amazon Kinesis Data Analytics for SQL Appli cations? DeepAR Forecasting Algorithm - Amazon SageMaker" + }, + { + "question": "A Machine Learning team runs its own training algor ithm on Amazon SageMaker. The training algorithm requires external assets. The team needs to submit both its own algorithm code and algorithmspecific parameters to Amazon SageMaker. What combination of services should the team use to build a custom algorithm in Amazon SageMaker? (Choose two.)", + "options": [ + "A. AWS Secrets Manager", + "B. AWS CodeStar", + "C. Amazon ECR", + "D. Amazon ECS" + ], + "correct": "", + "explanation": "The Machine Learning team wants to use its own trai ning algorithm on Amazon SageMaker, and submit both its own algorithm code and algorithm-sp ecific parameters. The best combination of services to build a custom algorithm in Amazon Sage Maker are Amazon ECR and Amazon S3. Amazon ECR is a fully managed container registry se rvice that allows you to store, manage, and deploy Docker container images. You can use Amazon ECR to create a Docker image that contains your training algorithm code and any dependencies o r libraries that it requires. You can also use Amazon ECR to push, pull, and manage your Docker im ages securely and reliably. Amazon S3 is a durable, scalable, and secure object storage service that can store any amount and type of data. You can use Amazon S3 to store your t raining data, model artifacts, and algorithmspecifi c parameters. You can also use Amazon S3 to access yo ur data and parameters from your training algorithm code, and to write your model ou tput to a specified location. Therefore, the Machine Learning team can use the fo llowing steps to build a custom algorithm in Amazon SageMaker: Write the training algorithm code in Python, using the Amazon SageMaker Python SDK or the Amazon SageMaker Containers library to interact wit h the Amazon SageMaker service. The code should be able to read the input data and parameter s from Amazon S3, and write the model output to Amazon S3. Create a Dockerfile that defines the base image, th e dependencies, the environment variables, and the commands to run the training algorithm code. Th e Dockerfile should also expose the ports that Amazon SageMaker uses to communicate with the conta iner. Build the Docker image using the Dockerfile, and ta g it with a meaningful name and version. Push the Docker image to Amazon ECR, and note the r egistry path of the image. Upload the training data, model artifacts, and algo rithm-specific parameters to Amazon S3, and note the S3 URIs of the objects. Create an Amazon SageMaker training job, using the Amazon SageMaker Python SDK or the AWS CLI. Specify the registry path of the Docker image, the S3 URIs of the input and output data, the algorithm-specific parameters, and other configurat ion options, such as the instance type, the number of instances, the IAM role, and the hyperpar ameters. Monitor the status and logs of the training job, an d retrieve the model output from Amazon S3. Reference: Use Your Own Training Algorithms Amazon ECR - Amazon Web Services Amazon S3 - Amazon Web Services", + "references": "" + }, + { + "question": "A company uses a long short-term memory (LSTM) mode l to evaluate the risk factors of a particular energy sector. The model reviews multi-page text documents to analyze each sentence of the text and categorize it as either a potential risk or no risk. The model is no t performing well, even though the Data Scientist has experimented with many different network structures and tuned the corresponding hyperparameters. Which approach will provide the MAXIMUM performance boost?", + "options": [ + "A. Initialize the words by term frequency-inverse do cument frequency (TF-IDF) vectors pretrained on", + "B. Use gated recurrent units (GRUs) instead of LSTM and run the training process until the validation", + "C. Reduce the learning rate and run the training pro cess until the training loss stops decreasing.", + "D. Initialize the words by word2vec embeddings pretr ained on a large collection of news articles" + ], + "correct": "D. Initialize the words by word2vec embeddings pretr ained on a large collection of news articles", + "explanation": "Initializing the words by word2vec embeddings pretr ained on a large collection of news articles related to the energy sector will provide the maxim um performance boost for the LSTM model. Word2vec is a technique that learns distributed rep resentations of words based on their cooccurrence in a large corpus of text. These representations ca pture semantic and syntactic similarities between words, which can help the LSTM model better understand the meaning and context of the sentences in the text documents. Usi ng word2vec embeddings that are pretrained on a relevant domain (energy sector) can further impro ve the performance by reducing the vocabulary mismatch and increasing the coverage of the words i n the text documents. Reference: AWS Machine Learning Specialty Exam Guide AWS Machine Learning Training - Text Classification with TF-IDF, LSTM, BERT: a comparison of performance AWS Machine Learning Training - Machine Learning - Exam Preparation Path", + "references": "" + }, + { + "question": "A Machine Learning Specialist previously trained a logistic regression model using scikit-learn on a local machine, and the Specialist now wants to deploy it to production for inference only. What steps should be taken to ensure Amazon SageMak er can host a model that was trained locally?", + "options": [ + "A. Build the Docker image with the inference code. T ag the Docker image with the registry hostname", + "B. Serialize the trained model so the format is comp ressed for deployment. Tag the Docker image", + "D. Build the Docker image with the inference code. C onfigure Docker Hub and upload the image to" + ], + "correct": "A. Build the Docker image with the inference code. T ag the Docker image with the registry hostname", + "explanation": "To deploy a model that was trained locally to Amazo n SageMaker, the steps are: Build the Docker image with the inference code. The inference code should include the model loading, data preprocessing, prediction, and postpr ocessing logic. The Docker image should also include the dependencies and libraries required by the inference code and the model. Tag the Docker image with the registry hostname and upload it to Amazon ECR. Amazon ECR is a fully managed container registry that makes it easy to st ore, manage, and deploy container images. The registry hostname is the Amazon ECR registry URI fo r your account and Region. You can use the AWS CLI or the Amazon ECR console to tag and push the D ocker image to Amazon ECR. Create a SageMaker model entity that points to the Docker image in Amazon ECR and the model artifacts in Amazon S3. The model entity is a logic al representation of the model that contains the information needed to deploy the model for inferenc e. The model artifacts are the files generated by the model training process, such as the model param eters and weights. You can use the AWS CLI, the SageMaker Python SDK, or the SageMaker console to c reate the model entity. Create an endpoint configuration that specifies the instance type and number of instances to use for hosting the model. The endpoint configuration also defines the production variants, which are the different versions of the model that you want to de ploy. You can use the AWS CLI, the SageMaker Python SDK, or the SageMaker console to create the endpoint configuration. Create an endpoint that uses the endpoint configura tion to deploy the model. The endpoint is a web service that exposes an HTTP API for inference requ ests. You can use the AWS CLI, the SageMaker Python SDK, or the SageMaker console to create the endpoint.", + "references": "AWS Machine Learning Specialty Exam Guide AWS Machine Learning Training - Deploy a Model on A mazon SageMaker AWS Machine Learning Training - Use Your Own Infere nce Code with Amazon SageMaker Hosting Services" + }, + { + "question": "A trucking company is collecting live image data fr om its fleet of trucks across the globe. The data i s growing rapidly and approximately 100 GB of new dat a is generated every day. The company wants to explore machine learning uses cases while ensuri ng the data is only accessible to specific IAM users. Which storage option provides the most processing f lexibility and will allow access control with IAM?", + "options": [ + "A. Use a database, such as Amazon DynamoDB, to store the images, and set the IAM policies to", + "B. Use an Amazon S3-backed data lake to store the ra w images, and set up the permissions using", + "C. Setup up Amazon EMR with Hadoop Distributed File System (HDFS) to store the files, and restrict", + "D. Configure Amazon EFS with IAM policies to make th e data available to Amazon EC2 instances" + ], + "correct": "B. Use an Amazon S3-backed data lake to store the ra w images, and set up the permissions using", + "explanation": "The best storage option for the trucking company is to use an Amazon S3-backed data lake to store the raw images, and set up the permissions using bu cket policies. A data lake is a centralized repository that allows you to store all your struct ured and unstructured data at any scale. Amazon S3 is the ideal choice for building a data lake becaus e it offers high durability, scalability, availabil ity, and security. You can store any type of data in Amazon S3, such as images, videos, audio, text, etc. You can also use AWS services such as Amazon Rekognitio n, Amazon SageMaker, and Amazon EMR to analyze and process the data in the data lake. To e nsure the data is only accessible to specific IAM users, you can use bucket policies to grant or deny access to the S3 buckets based on the IAM users identity or role. Bucket policies are JSON document s that specify the permissions for the bucket and the objects in it. You can use conditions to restri ct access based on various factors, such as IP addr ess, time, source, etc. By using bucket policies, you ca n control who can access the data in the data lake and what actions they can perform on it.", + "references": "AWS Machine Learning Specialty Exam Guide AWS Machine Learning Training - Build a Data Lake F oundation with Amazon S3 AWS Machine Learning Training - Using Bucket Polici es and User Policies" + }, + { + "question": "A credit card company wants to build a credit scori ng model to help predict whether a new credit card applicant will default on a credit card payment. The company has collected data from a large number of sources with thousands of raw attributes. Early experiments to t rain a classification model revealed that many attributes are highly correlated, the large number of features slo ws down the training speed significantly, and that there are some overfitting issues. The Data Scientist on this project would like to sp eed up the model training time without losing a lot of information from the original dataset. Which feature engineering technique should the Data Scientist use to meet the objectives?", + "options": [ + "A. Run self-correlation on all features and remove h ighly correlated features", + "B. Normalize all numerical values to be between 0 an d 1", + "C. Use an autoencoder or principal component analysi s (PCA) to replace original features with new", + "D. Cluster raw data using k-means and use sample dat a from each cluster to build a new dataset", + "A. Gather more data using Amazon Mechanical Turk and then retrain", + "B. Train an anomaly detection model instead of an ML P", + "C. Train an XGBoost model instead of an MLP", + "D. Add class weights to the MLPs loss function and t hen retrain" + ], + "correct": "D. Add class weights to the MLPs loss function and t hen retrain", + "explanation": "The best technique to improve the recall of the MLP for the target class of interest is to add class weights to the MLPs loss function and then retrain. Class weights are a way of assigning different importance to each class in the dataset, such that the model will pay more attention to the classes with higher weights. This can help mitigate the cla ss imbalance problem, where the model tends to favor the majority class and ignore the minority cl ass. By increasing the weight of the target class o f interest, the model will try to reduce the false ne gatives and increase the true positives, which will improve the recall metric. Adding class weights to the loss function is also a quick and easy solution , as it does not require gathering more data, changin g the model architecture, or switching to a different algorithm.", + "references": "AWS Machine Learning Specialty Exam Guide AWS Machine Learning Training - Deep Learning with Amazon SageMaker AWS Machine Learning Training - Class Imbalance and Weighted Loss Functions" + }, + { + "question": "A Machine Learning Specialist works for a credit ca rd processing company and needs to predict which transactions may be fraudulent in near-real time. S pecifically, the Specialist must train a model that returns the probability that a given transaction may fraudulent . How should the Specialist frame this business probl em?", + "options": [ + "A. Streaming classification", + "B. Binary classification", + "C. Multi-category classification", + "D. Regression classification" + ], + "correct": "B. Binary classification", + "explanation": "The business problem of predicting whether a new cr edit card applicant will default on a credit card payment can be framed as a binary classification pr oblem. Binary classification is the task of predicting a discrete class label output for an exa mple, where the class label can only take one of tw o possible values. In this case, the class label can be either oedefault or oeno default , indicating whether the applicant will or will not default on a credit card payment. A binary classification model can return the probability that a given applicant b elongs to each class, and then assign the applicantto the class with the highest probability. For exam ple, if the model predicts that an applicant has a 0.8 probability of defaulting and a 0.2 probability of not defaulting, then the model will classify th e applicant as oedefault . Binary classification is s uitable for this problem because the outcome of interest is categorical and binary, and the model n eeds to return the probability of each outcome.", + "references": "AWS Machine Learning Specialty Exam Guide AWS Machine Learning Training - Classification vs R egression in Machine Learning" + }, + { + "question": "A real estate company wants to create a machine lea rning model for predicting housing prices based on a historical dataset. The dataset contains 32 feature s. Which model will meet the business requirement?", + "options": [ + "A. Logistic regression", + "B. Linear regression", + "C. K-means", + "D. Principal component analysis (PCA)" + ], + "correct": "B. Linear regression", + "explanation": "The best model for predicting housing prices based on a historical dataset with 32 features is linear regression. Linear regression is a supervised learn ing algorithm that fits a linear relationship betwe en a dependent variable (housing price) and one or mor e independent variables (features). Linear regression can handle multiple features and output a continuous value for the housing price. Linear regression can also return the coefficients of the features, which indicate how each feature affects the housing price. Linear regression is suitable fo r this problem because the outcome of interest is numerical and continuous, and the model needs to ca pture the linear relationship between the features and the outcome.", + "references": "AWS Machine Learning Specialty Exam Guide AWS Machine Learning Training - Regression vs Class ification in Machine Learning AWS Machine Learning Training - Linear Regression w ith Amazon SageMaker" + }, + { + "question": "A Machine Learning Specialist wants to bring a cust om algorithm to Amazon SageMaker. The Specialist implements the algorithm in a Docker container supp orted by Amazon SageMaker. How should the Specialist package the Docker contai ner so that Amazon SageMaker can launch the training correctly?", + "options": [ + "A. Modify the bash_profile file in the container and add a bash command to start the training", + "B. Use CMD config in the Dockerfile to add the train ing program as a CMD of the image", + "C. Configure the training program as an ENTRYPOINT n amed train", + "D. Copy the training program to directory /opt/ml/tr ain" + ], + "correct": "C. Configure the training program as an ENTRYPOINT n amed train", + "explanation": "To use a custom algorithm in Amazon SageMaker, the Docker container image must have an executable file named train that acts as the ENTRYP OINT for the container. This file is responsible fo r running the training code and communicating with th e Amazon SageMaker service. The train file must be in the PATH of the container and have execu te permissions. The other options are not valid ways to package the Docker container for Amazon Sag eMaker. Reference: Use Docker containers to build models - Amazon Sage Maker Create a container with your own algorithms and mod els - Amazon SageMaker", + "references": "" + }, + { + "question": "A Data Scientist needs to analyze employment dat", + "options": [ + "A. The dataset contains approximately 10 million", + "B. Cross-validation", + "C. Numerical value binning", + "D. High-degree polynomial transformation" + ], + "correct": "", + "explanation": "To fix the incorrectly skewed data, the Data Scient ist can apply two feature transformations: numerical value binning and logarithmic transformat ion. Numerical value binning is a technique that groups continuous values into discrete bins or cate gories. This can help reduce the skewness of the data by creating more balanced frequency distributi ons. Logarithmic transformation is a technique that applies the natural logarithm function to each value in the data. This can help reduce the right skewness of the data by compressing the large value s and expanding the small values. Both of these transformations can make the data more suitable for machine learning algorithms that assume normality of the data. Reference: Data Transformation - Amazon SageMaker Transforming Skewed Data for Machine Learning", + "references": "" + }, + { + "question": "A Machine Learning Specialist is given a structured dataset on the shopping habits of a companys customer base. The dataset contains thousands of columns of data and hundreds of numerical columns for each customer. The Specialist wants to identify whether there are natural groupings for these columns across all customers and visualize the results as quickly as p ossible. What approach should the Specialist take to accompl ish these tasks?", + "options": [ + "A. Embed the numerical features using the t-distribu ted stochastic neighbor embedding (t-SNE)", + "B. Run k-means using the Euclidean distance measure for different values of k and create an elbow", + "D. Run k-means using the Euclidean distance measure for different values of k and create box plots" + ], + "correct": "A. Embed the numerical features using the t-distribu ted stochastic neighbor embedding (t-SNE)", + "explanation": "The best approach to identify and visualize the nat ural groupings for the numerical columns across all customers is to embed the numerical features using the t-distributed stochastic neighbor embedding (t-SNE) algorithm and create a scatter plot. t-SNE is a dimensionality reduction technique that can project high-dimensional data into a lower-dimensio nal space, while preserving the local structure and distances of the data points. A scatter plot ca n then show the clusters of data points in the reduced space, where each point represents a custom er and the color indicates the cluster membership. This approach can help the Specialist q uickly explore the patterns and similarities among the customers based on their numerical featur es. The other options are not as effective or efficient as the t-SNE approach. Running k-means for different values of k and creating an elbow plot ca n help determine the optimal number of clusters, but it does not provide a visual representation of the clusters or the customers. Embedding the numerical features using t-SNE and creating a line graph does not make sense, as a line graph is used to show the change of a variable over time, not the distribution of data points in a space. Running km eans for different values of k and creating box plots fo r each numerical column within each cluster can provide some insights into the statistics of ea ch cluster, but it is very time-consuming and cumbersome to create and compare thousands of box p lots. Reference: Dimensionality Reduction - Amazon SageMaker Visualize high dimensional data using t-SNE - Amazo n SageMaker", + "references": "" + }, + { + "question": "A Machine Learning Specialist is planning to create a long-running Amazon EMR cluster. The EMR cluster will have 1 master node, 10 core nodes, and 20 task node s. To save on costs, the Specialist will use Spot Instances in the EMR cluster. Which nodes should the Specialist launch on Spot In stances?", + "options": [ + "A. Master node", + "B. Any of the core nodes", + "C. Any of the task nodes", + "D. Both core and task nodes", + "A. The target", + "B. Plot a histogram of the features and compute thei r standard deviation. Remove features with high", + "C. Plot a histogram of the features and compute thei r standard deviation. Remove features with low", + "D. Build a heatmap showing the correlation of the da taset against itself. Remove features with low" + ], + "correct": "D. Build a heatmap showing the correlation of the da taset against itself. Remove features with low", + "explanation": "Feature selection is the process of reducing the nu mber of input variables to those that are most relevant for predicting the target variable. One wa y to do this is to run a correlation check of all features against the target variable and remove fea tures with low target variable correlation scores. This means that these features have little or no li near relationship with the target variable and are not useful for the prediction. This can reduce the models complexity and improve its performance. Reference: Feature engineering - Machine Learning Lens Feature Selection For Machine Learning in Python", + "references": "" + }, + { + "question": "A health care company is planning to use neural net works to classify their X-ray images into normal and abnormal classes. The labeled data is divided i nto a training set of 1,000 images and a test set o f 200 images. The initial training of a neural networ k model with 50 hidden layers yielded 99% accuracy on the training set, but only 55% accuracy on the test set. What changes should the Specialist consider to solv e this issue? (Choose three.)", + "options": [ + "A. Choose a higher number of layers", + "B. Choose a lower number of layers", + "C. Choose a smaller learning rate", + "D. Enable dropout" + ], + "correct": "", + "explanation": "Explanation/Reference: The problem described in the question is a case of overfitting, where the neural network model performs well on the training data but poorly on th e test data. This means that the model has learned the noise and specific patterns of the trai ning data, but cannot generalize to new and unseen data. To solve this issue, the Specialist should co nsider the following changes: Choose a lower number of layers: Reducing the numbe r of layers can reduce the complexity and capacity of the neural network model, making it les s prone to overfitting. A model with 50 hidden layers is likely too deep for the given data size a nd task. A simpler model with fewer layers can lear n the essential features of the data without memorizi ng the noise. Enable dropout: Dropout is a regularization techniq ue that randomly drops out some units in the neural network during training. This prevents the u nits from co-adapting too much and forces the model to learn more robust features. Dropout can im prove the generalization and test performance of the model by reducing overfitting. Enable early stopping: Early stopping is another re gularization technique that monitors the validation error during training and stops the training proces s when the validation error stops decreasing or starts increasing. This prevents the model from ove rtraining on the training data and reduces overfitting.", + "references": "Deep Learning - Machine Learning Lens How to Avoid Overfitting in Deep Learning Neural Ne tworks How to Identify Overfitting Machine Learning Models in Scikit-Learn" + }, + { + "question": "A Machine Learning Specialist is attempting to buil d a linear regression model. Given the displayed residual plot only, what is the MOST likely problem with the model?", + "options": [ + "A. Linear regression is inappropriate. The residuals do not have constant variance.", + "B. Linear regression is inappropriate. The underlyin g data has outliers.", + "C. Linear regression is appropriate. The residuals h ave a zero mean.", + "D. Linear regression is appropriate. The residuals h ave constant variance." + ], + "correct": "A. Linear regression is inappropriate. The residuals do not have constant variance.", + "explanation": "A residual plot is a type of plot that displays the values of a predictor variable in a regression mod el along the x-axis and the values of the residuals al ong the y-axis. This plot is used to assess whether or not the residuals in a regression model are normall y distributed and whether or not they exhibit heteroscedasticity. Heteroscedasticity means that t he variance of the residuals is not constant across different values of the predictor variable. This vi olates one of the assumptions of linear regression and can lead to biased estimates and unreliable pre dictions. The displayed residual plot shows a clear pattern of heteroscedasticity, as the residua ls spread out as the fitted values increase. This indicates that linear regression is inappropriate f or this data and a different model should be used. Reference: Regression - Amazon Machine Learning How to Create a Residual Plot by Hand How to Create a Residual Plot in Python", + "references": "" + }, + { + "question": "A machine learning specialist works for a fruit pro cessing company and needs to build a system that categorizes apples into three types. The specialist has collected a dataset that contains 150 images for each type of apple and applied transfer learnin g on a neural network that was pretrained on ImageNet with this dataset. The company requires at least 85% accuracy to make use of the model. After an exhaustive grid search, the optimal hyperp arameters produced the following: 68% accuracy on the training set 67% accuracy on the validation set What can the machine learning specialist do to impr ove the systems accuracy?", + "options": [ + "A. Upload the model to an Amazon SageMaker notebook instance and use the Amazon SageMaker", + "B. Add more data to the training set and retrain the model using transfer learning to reduce the bias.", + "C. Use a neural network model with more layers that are pretrained on ImageNet and apply transfer", + "D. Train a new model using the current neural networ k architecture." + ], + "correct": "B. Add more data to the training set and retrain the model using transfer learning to reduce the bias.", + "explanation": "The problem described in the question is a case of underfitting, where the neural network model performs poorly on both the training and validation sets. This means that the model has not learned the features of the data well enough and has high b ias. To solve this issue, the machine learning specialist should consider the following change: Add more data to the training set and retrain the m odel using transfer learning to reduce the bias: Adding more data to the training set can help the m odel learn more patterns and variations in the data and improve its performance. Transfer learning can also help the model leverage the knowledge from the pre-trained network and adapt it to the ne w data. This can reduce the bias and increase the accuracy of the model.", + "references": "Transfer learning for TensorFlow image classificati on models in Amazon SageMaker Transfer learning for custom labels using a TensorF low container and oebring your own algorithm in Amazon SageMaker Machine Learning Concepts - AWS Training and Certif ication" + }, + { + "question": "A company uses camera images of the tops of items d isplayed on store shelves to determine which items were removed and which ones still remain. After sev eral hours of data labeling, the company has a total of 1,000 hand-labeled images covering 10 distinct item s. The training results were poor. Which machine learning approach fulfills the compan ys long-term needs?", + "options": [ + "A. Convert the images to grayscale and retrain the m odel", + "B. Reduce the number of distinct items from 10 to 2, build the model, and iterate", + "C. Attach different colored labels to each item, tak e the images again, and build the model", + "D. Augment training data for each item using image v ariants like inversions and translations, build" + ], + "correct": "D. Augment training data for each item using image v ariants like inversions and translations, build", + "explanation": "Data augmentation is a technique that can increase the size and diversity of the training data by applying various transformations to the original im ages, such as inversions, translations, rotations, scaling, cropping, flipping, and color variations. Data augmentation can help improve the performance and generalization of image classificat ion models by reducing overfitting and introducing more variability to the data. Data augm entation is especially useful when the original data is limited or imbalanced, as in the case of th e companys problem. By augmenting the training data for each item using image variants, the compan y can build a more robust and accurate model that can recognize the items on the store shelves f rom different angles, positions, and lighting conditions. The company can also iterate on the mod el by adding more data or fine-tuning the hyperparameters to achieve better results.", + "references": "Build high performing image classification models u sing Amazon SageMaker JumpStart The Effectiveness of Data Augmentation in Image Cla ssification using Deep Learning Data augmentation for improving deep learning in im age classification problem Class-Adaptive Data Augmentation for Image Classifi cation" + }, + { + "question": "A Data Scientist is developing a binary classifier to predict whether a patient has a particular disea se on a series of test results. The Data Scientist has data on 400 patients randomly selected from the population. The disease is seen in 3% of the popula tion. Which cross-validation strategy should the Data Sci entist adopt?", + "options": [ + "A. A k-fold cross-validation strategy with k=5", + "B. A stratified k-fold cross-validation strategy wit h k=5", + "C. A k-fold cross-validation strategy with k=5 and 3 repeats", + "D. An 80 stratified split between training and valid ation" + ], + "correct": "B. A stratified k-fold cross-validation strategy wit h k=5", + "explanation": "A stratified k-fold cross-validation strategy is a technique that preserves the class distribution in each fold. This is important for imbalanced datasets, su ch as the one in the question, where the disease is seen in only 3% of the population. If a random k-fo ld cross-validation strategy is used, some folds may have no positive cases or very few, which would lead to poor estimates of the model performance. A stratified k-fold cross-validation s trategy ensures that each fold has the same proportion of positive and negative cases as the wh ole dataset, which makes the evaluation more reliable and robust. A k-fold cross-validation stra tegy with k=5 and 3 repeats is also a possible opti on, but it is more computationally expensive and may no t be necessary if the stratification is done properly. An 80 stratified split between training a nd validation is another option, but it uses less data for training and validation than k-fold cross- validation, which may result in higher variance and lower accuracy of the estimates. Reference: AWS Machine Learning Specialty Certification Exam G uide AWS Machine Learning Training: Model Evaluation How to Fix k-Fold Cross-Validation for Imbalanced C lassification", + "references": "" + }, + { + "question": "A technology startup is using complex deep neural n etworks and GPU compute to recommend the companys products to its existing customers based u pon each customers habits and interactions. The solution currently pulls each dataset from an A mazon S3 bucket before loading the data into a TensorFlow model pulled from the companys Git repos itory that runs locally. This job then runs for several hours while continually outputting its prog ress to the same S3 bucket. The job can be paused, restarted, and continued at any time in the event o f a failure, and is run from a central queue. Senior managers are concerned about the complexity of the solutions resource management and the costs involved in repeating the process regular ly. They ask for the workload to be automated so it runs once a week, starting Monday and completing by the close of business Friday. Which architecture should be used to scale the solu tion at the lowest cost?", + "options": [ + "A. Implement the solution using AWS Deep Learning Co ntainers and run the container as a job using", + "B. Implement the solution using a low-cost GPU-compa tible Amazon EC2 instance and use the AWS", + "C. Implement the solution using AWS Deep Learning Co ntainers, run the workload using AWS Fargate" + ], + "correct": "A. Implement the solution using AWS Deep Learning Co ntainers and run the container as a job using", + "explanation": "The best architecture to scale the solution at the lowest cost is to implement the solution using AWS Deep Learning Containers and run the container as a job using AWS Batch on a GPU-compatible Spot Instance. This option has the following advantages: AWS Deep Learning Containers: These are Docker imag es that are pre-installed and optimized with popular deep learning frameworks such as TensorFlow , PyTorch, and MXNet. They can be easily deployed on Amazon EC2, Amazon ECS, Amazon EKS, and AWS Fargate. They can also be integrated with AWS Batch to run containerized batch jobs. Usi ng AWS Deep Learning Containers can simplify the setup and configuration of the deep learning en vironment and reduce the complexity of the resource management. AWS Batch: This is a fully managed service that ena bles you to run batch computing workloads on AWS. You can define compute environments, job queue s, and job definitions to run your batch jobs. You can also use AWS Batch to automatically provisi on compute resources based on the requirements of the batch jobs. You can specify the type and quantity of the compute resources, such as GPU instances, and the maximum price you are wil ling to pay for them. You can also use AWS Batch to monitor the status and progress of your ba tch jobs and handle any failures or interruptions. GPU-compatible Spot Instance: This is an Amazon EC2 instance that uses a spare compute capacity that is available at a lower price than the On-Dema nd price. You can use Spot Instances to run your deep learning training jobs at a lower cost, as lon g as you are flexible about when your instances run and how long they run. You can also use Spot Instan ces with AWS Batch to automatically launch and terminate instances based on the availability and p rice of the Spot capacity. You can also use Spot Instances with Amazon EBS volumes to store your dat asets, checkpoints, and logs, and attach them to your instances when they are launched. This way, you can preserve your data and resume your training even if your instances are interrupted.", + "references": "AWS Deep Learning Containers AWS Batch Amazon EC2 Spot Instances Using Amazon EBS Volumes with Amazon EC2 Spot Insta nces" + }, + { + "question": "A media company with a very large archive of unlabe led images, text, audio, and video footage wishes to index its assets to allow rapid identific ation of relevant content by the Research team. The company wants to use machine learning to accelerate the efforts of its in-house researchers who have limited machine learning expertise. Which is the FASTEST route to index the assets?", + "options": [ + "A. Use Amazon Rekognition, Amazon Comprehend, and Am azon Transcribe to tag data into distinct", + "B. Create a set of Amazon Mechanical Turk Human Inte lligence Tasks to label all footage.", + "C. Use Amazon Transcribe to convert speech to text. Use the Amazon SageMaker Neural Topic Model", + "D. Use the AWS Deep Learning AMI and Amazon EC2 GPU instances to create custom models for" + ], + "correct": "A. Use Amazon Rekognition, Amazon Comprehend, and Am azon Transcribe to tag data into distinct", + "explanation": "Amazon Rekognition, Amazon Comprehend, and Amazon T ranscribe are AWS machine learning services that can analyze and extract metadata from images, text, audio, and video content. These services are easy to use, scalable, and do not requ ire any machine learning expertise. They can help the media company to quickly index its assets and e nable rapid identification of relevant content by the research team. Using these services is the fast est route to index the assets, compared to the other options that involve human intervention, cust om model development, or additional steps. Reference: AWS Media Intelligence Solutions AWS Machine Learning Services The Best Services For Running Machine Learning Mode ls On AWS", + "references": "" + }, + { + "question": "A Machine Learning Specialist is working for an onl ine retailer that wants to run analytics on every customer visit, processed through a machine learnin g pipeline. The data needs to be ingested by Amazon Kinesis Data Streams at up to 100 transactio ns per second, and the JSON data blob is 100 KB in size. What is the MINIMUM number of shards in Kinesis Dat a Streams the Specialist should use to successfully ingest this data?", + "options": [ + "A. 1 shards", + "B. 10 shards", + "C. 100 shards", + "D. 1,000 shards" + ], + "correct": "A. 1 shards", + "explanation": "According to the Amazon Kinesis Data Streams docume ntation, the maximum size of data blob (the data payload before Base64-encoding) per record is 1 MB. The maximum number of records that can be sent to a shard per second is 1,000. Therefore, the maximum throughput of a shard is 1 MB/sec for input and 2 MB/sec for output. In this case, th e input throughput is 100 transactions per second * 100 KB per transaction = 10 MB/sec. Therefore, the minimum number of shards required is 10 MB/sec / 1 MB/sec = 10 shards. However, the questio n asks for the minimum number of shards in Kinesis Data Streams, not the minimum number of sha rds per stream. A Kinesis Data Streams account can have multiple streams, each with its ow n number of shards. Therefore, the minimum number of shards in Kinesis Data Streams is 1, whic h is the minimum number of shards per stream. Reference: Amazon Kinesis Data Streams Terminology and Concept s Amazon Kinesis Data Streams Limits", + "references": "" + }, + { + "question": "A Machine Learning Specialist is deciding between b uilding a naive Bayesian model or a full Bayesian network for a classification problem. The Specialis t computes the Pearson correlation coefficients between each feature and finds that their absolute values range between 0.1 to 0.95. Which model describes the underlying data in this s ituation?", + "options": [ + "A. A naive Bayesian model, since the features are al l conditionally independent.", + "B. A full Bayesian network, since the features are a ll conditionally independent.", + "C. A naive Bayesian model, since some of the feature s are statistically dependent.", + "D. A full Bayesian network, since some of the featur es are statistically dependent." + ], + "correct": "D. A full Bayesian network, since some of the featur es are statistically dependent.", + "explanation": "A naive Bayesian model assumes that the features ar e conditionally independent given the class label. This means that the joint probability of the features and the class can be factorized as the product of the class prior and the feature likeliho ods. A full Bayesian network, on the other hand, does not make this assumption and allows for modeli ng arbitrary dependencies between the features and the class using a directed acyclic gra ph. In this case, the joint probability of the feat ures and the class is given by the product of the condit ional probabilities of each node given its parents in the graph. If the features are statistically depend ent, meaning that their correlation coefficients ar e not close to zero, then a naive Bayesian model woul d not capture these dependencies and would likely perform worse than a full Bayesian network t hat can account for them. Therefore, a full Bayesian network describes the underlying data bett er in this situation. Reference: Naive Bayes and Text Classification I Bayesian Networks", + "references": "" + }, + { + "question": "A Data Scientist is building a linear regression mo del and will use resulting p-values to evaluate the statistical significance of each coefficient. Upon inspection of the dataset, the Data Scientist disco vers that most of the features are normally distributed. The plot of one feature in the dataset is shown in the graphic. What transformation should the Data Scientist apply to satisfy the statistical assumptions of the linear regression model?", + "options": [ + "A. Exponential transformation", + "B. Logarithmic transformation", + "C. Polynomial transformation", + "D. Sinusoidal transformation Correct Answer: B" + ], + "correct": "", + "explanation": "The plot in the graphic shows a right-skewed distri bution, which violates the assumption of normality for linear regression. To correct this, t he Data Scientist should apply a logarithmic transformation to the feature. This will help to ma ke the distribution more symmetric and closer to a normal distribution, which is a key assumption for linear regression. Reference: Linear Regression Linear Regression with Amazon Machine Learning Machine Learning on AWS", + "references": "" + }, + { + "question": "A Machine Learning Specialist is assigned to a Frau d Detection team and must tune an XGBoost model, which is working appropriately for test dat", + "options": [ + "A. However, with unknown data, it is not working as expected. The existing parameters are provided", + "B. Increase the max_depth parameter value.", + "C. Lower the max_depth parameter value.", + "D. Update the objective to binary:logistic." + ], + "correct": "B. Increase the max_depth parameter value.", + "explanation": "Overfitting occurs when a model performs well on th e training data but poorly on the test data. This is often because the model has learned the training data too well and is not able to generalize to new data. To avoid overfitting, the Machine Learning Sp ecialist should lower the max_depth parameter value. This will reduce the complexity of the model and make it less likely to overfit. According to t he XGBoost documentation1, the max_depth parameter con trols the maximum depth of a tree and lower values can help prevent overfitting. The docu mentation also suggests other ways to control overfitting, such as adding randomness, using regul arization, and using early stopping1. Reference: XGBoost Parameters", + "references": "" + }, + { + "question": "A data scientist is developing a pipeline to ingest streaming web traffic dat A. The data scientist needs to implement a process t o identify unusual web traffic patterns as part of the pipeline. The patterns will be used downstream for alerting and incident response. The data scientist has access to unlabeled historic data to use, if needed. The solution needs to do the following: Calculate an anomaly score for each web traffic ent ry. Adapt unusual event identification to changing web patterns over time. Which approach should the data scientist implement to meet these requirements?", + "options": [ + "B. Use historic web traffic data to train an anomaly detection model using the Amazon SageMaker", + "C. Use historic web traffic data to train an anomaly detection model using the Amazon SageMaker", + "D. Collect the streaming data using Amazon Kinesis D ata Firehose. Map the delivery stream as an" + ], + "correct": "D. Collect the streaming data using Amazon Kinesis D ata Firehose. Map the delivery stream as an", + "explanation": "Amazon Kinesis Data Analytics is a service that all ows users to analyze streaming data in real time using SQL queries. Amazon Random Cut Forest (RCF) i s a SQL extension that enables anomaly detection on streaming data. RCF is an unsupervised machine learning algorithm that assigns an anomaly score to each data point based on how diffe rent it is from the rest of the data. A sliding window is a type of window that moves along with th e data stream, so that the anomaly detection model can adapt to changing patterns over time. A t umbling window is a type of window that has a fixed size and does not overlap with other windows, so that the anomaly detection model is based on a fixed period of time. Therefore, option D is t he best approach to meet the requirements of the question, as it uses RCF to calculate anomaly score s for each web traffic entry and uses a sliding window to adapt to changing web patterns over time. Option A is incorrect because Amazon SageMaker Rand om Cut Forest (RCF) is a built-in model that can be used to train and deploy anomaly detection m odels on batch or streaming data, but it requires more steps and resources than using the RC F SQL extension in Amazon Kinesis Data Analytics. Option B is incorrect because Amazon Sag eMaker XGBoost is a built-in model that can be used for supervised learning tasks such as classifi cation and regression, but not for unsupervised learning tasks such as anomaly detection. Option C is incorrect because k-Nearest Neighbors (kNN) is a SQL extension that can be used for classification and regression tasks on streaming data, but not fo r anomaly detection. Moreover, using a tumbling windo w would not allow the anomaly detection model to adapt to changing web patterns over time.", + "references": "Using CloudWatch anomaly detection Anomaly Detection With CloudWatch Performing Real-time Anomaly Detection using AWS What Is AWS Anomaly Detection? (And Is There A Bett er Option?)" + }, + { + "question": "A Data Scientist received a set of insurance record s, each consisting of a record ID, the final outcom e among 200 categories, and the date of the final out come. Some partial information on claim contents is also provided, but only for a few of th e 200 categories. For each outcome category, there are hundreds of records distributed over the past 3 years. The Data Scientist wants to predict how many claims to expect in each category from month t o month, a few months in advance. What type of machine learning model should be used?", + "options": [ + "A. Classification month-to-month using supervised le arning of the 200 categories based on claim", + "B. Reinforcement learning using claim IDs and timest amps where the agent will identify how many", + "C. Forecasting using claim IDs and timestamps to ide ntify how many claims in each category to", + "D. Classification with supervised learning of the ca tegories for which partial information on claim" + ], + "correct": "C. Forecasting using claim IDs and timestamps to ide ntify how many claims in each category to", + "explanation": ": Forecasting is a type of machine learning model t hat predicts future values of a target variable based on historical data and other features. Foreca sting is suitable for problems that involve timeser ies data, such as the number of claims in each category from month to month. Forecasting can handle multiple categories of the target variable, as well as missing or partial information on some features. Therefore, option C is the best choice fo r the given problem. Option A is incorrect because classification is a t ype of machine learning model that assigns a label to an input based on predefined categories. Classifica tion is not suitable for predicting continuous or numerical values, such as the number of claims in e ach category from month to month. Moreover, classification requires sufficient and complete inf ormation on the features that are relevant to the target variable, which is not the case for the give n problem. Option B is incorrect because reinforcement learning is a type of machine learnin g model that learns from its own actions and rewards in an interactive environment. Reinforcemen t learning is not suitable for problems that involve historical data and do not require an agent to take actions. Option D is incorrect because it combines two different types of machine learning mo dels, which is unnecessary and inefficient. Moreover, classification is not suitable for predic ting the number of claims in some categories, as explained in option A.", + "references": "Forecasting | AWS Solutions for Machine Learning (A I/ML) | AWS Solutions Library Time Series Forecasting Service \" Amazon Forecast \" Amazon Web Services Amazon Forecast: Guide to Predicting Future Outcome s - Onica Amazon Launches What-If Analyses for Machine Learni ng Forecasting \u00a6" + }, + { + "question": "A company that promotes healthy sleep patterns by p roviding cloud-connected devices currently hosts a sleep tracking application on AWS. The appl ication collects device usage information from device users. The company's Data Science team is bu ilding a machine learning model to predict if and when a user will stop utilizing the company's devic es. Predictions from this model are used by a downstream application that determines the best app roach for contacting users. The Data Science team is building multiple versions of the machine learning model to evaluate each version against the companys business goals. To mea sure long-term effectiveness, the team wants to run multiple versions of the model in parallel f or long periods of time, with the ability to contro l the portion of inferences served by the models. Which solution satisfies these requirements with MI NIMAL effort?", + "options": [ + "A. Build and host multiple models in Amazon SageMake r. Create multiple Amazon SageMaker", + "B. Build and host multiple models in Amazon SageMake r. Create an Amazon SageMaker endpoint configuration with multiple production variants. Pr ogrammatically control the portion of the", + "C. Build and host multiple models in Amazon SageMake r Neo to take into account different types of", + "D. Build and host multiple models in Amazon SageMake r. Create a single endpoint that accesses" + ], + "correct": "B. Build and host multiple models in Amazon SageMake r. Create an Amazon SageMaker endpoint configuration with multiple production variants. Pr ogrammatically control the portion of the", + "explanation": "Amazon SageMaker is a service that allows users to build, train, and deploy ML models on AWS. Amazon SageMaker endpoints are scalable and secure web services that can be used to perform real-time inference on ML models. An endpoint confi guration defines the models that are deployed and the resources that are used by the endpoint. An endpoint configuration can have multiple production variants, each representing a different version or variant of a model. Users can specify th e portion of the inferences served by each production variant using the initialVariantWeight parameter. Users can also programmatically update the endpoint configuration to change the portion of the inferences served by each production variant using the UpdateEndpointWeightsAndCapacities API. Therefore, option B is the best solution to satisfy the requirements with minimal effort. Option A is incorrect because creating multiple end points for each model would incur more cost and complexity than using a single endpoint with multip le production variants. Moreover, controlling the invocation of different models at the application l ayer would require more custom logic and coordination than using the UpdateEndpointWeightsAn dCapacities API. Option C is incorrect because Amazon SageMaker Neo is a service that allo ws users to optimize ML models for different hardware platforms, such as edge devices. It is not relevant to the problem of running multiple versions of a model in parallel for long periods of time. Option D is incorrect because Amazon SageMaker batch transform is a service that allows users to perform asynchronous inference on large datasets. It is not suitable for the problem of per forming real-time inference on streaming data from device users.", + "references": "Deploying models to Amazon SageMaker hosting servic es - Amazon SageMaker Update an Amazon SageMaker endpoint to accommodate new models - Amazon SageMaker UpdateEndpointWeightsAndCapacities - Amazon SageMak er" + }, + { + "question": "An agricultural company is interested in using mach ine learning to detect specific types of weeds in a 100-acre grassland field. Currently, the company us es tractor-mounted cameras to capture multiple images of the field as 10 \u00c3-- 10 grids. The company also has a large training dataset that consists of annotated images of popular weed classes like broad leaf and non-broadleaf docks. The company wants to build a weed detection model t hat will detect specific types of weeds and the location of each type within the field. Once the mo del is ready, it will be hosted on Amazon SageMaker endpoints. The model will perform real-ti me inferencing using the images captured by the cameras. Which approach should a Machine Learning Specialist take to obtain accurate predictions?", + "options": [ + "A. Prepare the images in RecordIO format and upload them to Amazon S3. Use Amazon SageMaker", + "B. Prepare the images in Apache Parquet format and u pload them to Amazon S3. Use Amazon", + "C. Prepare the images in RecordIO format and upload them to Amazon S3. Use Amazon SageMaker", + "D. Prepare the images in Apache Parquet format and u pload them to Amazon S3. Use Amazon" + ], + "correct": "C. Prepare the images in RecordIO format and upload them to Amazon S3. Use Amazon SageMaker", + "explanation": "The problem of detecting specific types of weeds an d their location within the field is an example of object detection, which is a type of machine learni ng model that identifies and localizes objects in a n image. Amazon SageMaker provides a built-in object detection algorithm that uses a single-shot multibox detector (SSD) to perform real-time infere nce on streaming images. The SSD algorithm can handle multiple objects of varying sizes and scales in an image, and generate bounding boxes and scores for each object category. Therefore, option C is the best approach to obtain accurate predictions. Option A is incorrect because image classification is a type of machine learning model that assigns a label to an image based on predefined categories. I mage classification is not suitable for localizing objects within an image, as it does not provide bou nding boxes or scores for each object. Option B is incorrect because Apache Parquet is a columnar stor age format that is optimized for analytical queries. Apache Parquet is not suitable for storing images, as it does not preserve the spatial information of the pixels. Option D is incorrect be cause it combines the wrong format (Apache Parquet) and the wrong algorithm (image classificat ion) for the given problem, as explained in options A and B.", + "references": "Object Detection algorithm now available in Amazon SageMaker Image classification and object detection using Ama zon Rekognition Custom Labels and Amazon SageMaker JumpStart Object Detection with Amazon SageMaker - W3Schools aws-samples/amazon-sagemaker-tensorflow-object-dete ction-api" + }, + { + "question": "A manufacturer is operating a large number of facto ries with a complex supply chain relationship where unexpected downtime of a machine can cause pr oduction to stop at several factories. A data scientist wants to analyze sensor data from the fac tories to identify equipment in need of preemptive maintenance and then dispatch a service team to prevent unplanned downtime. The sensor readings from a single machine can include u p to 200 data points including temperatures, voltages, vibrations, RPMs, and pressure readings. To collect this sensor data, the manufacturer deplo yed Wi-Fi and LANs across the factories. Even though many factory locations do not have reliable or high-speed internet connectivity, the manufacturer would like to maintain near-real-time inference capabilities. Which deployment architecture for the model will ad dress these business requirements?", + "options": [ + "A. Deploy the model in Amazon SageMaker. Run sensor data through this model to predict which", + "B. Deploy the model on AWS IoT Greengrass in each fa ctory. Run sensor data through this model to", + "C. Deploy the model to an Amazon SageMaker batch tra nsformation job. Generate inferences in a", + "D. Deploy the model in Amazon SageMaker and use an I oT rule to write data to an Amazon", + "A. Moreover, this option would introduce" + ], + "correct": "B. Deploy the model on AWS IoT Greengrass in each fa ctory. Run sensor data through this model to", + "explanation": "AWS IoT Greengrass is a service that extends AWS to edge devices, such as sensors and machines, so they can act locally on the data they generate, whi le still using the cloud for management, analytics, and durable storage. AWS IoT Greengrass enables loc al device messaging, secure data transfer, and local computing using AWS Lambda functions and mach ine learning models. AWS IoT Greengrass can run machine learning inference locally on devices u sing models that are created and trained in the cloud. This allows devices to respond quickly to lo cal events, even when they are offline or have intermittent connectivity. Therefore, option B is t he best deployment architecture for the model to address the business requirements of the manufactur er. Option A is incorrect because deploying the model i n Amazon SageMaker would require sending the sensor data to the cloud for inference, which would not work well for factory locations that do not have reliable or high-speed internet connectivity. Moreover, this option would not provide near-realti me inference capabilities, as there would be latency a nd bandwidth issues involved in transferring the data to and from the cloud. Option C is incorre ct because deploying the model to an Amazon SageMaker batch transformation job would not provid e near-real-time inference capabilities, as batch transformation is an asynchronous process tha t operates on large datasets. Batch transformation is not suitable for streaming data t hat requires low-latency responses. Option D is incorrect because deploying the model in Amazon Sag eMaker and using an IoT rule to write data to an Amazon DynamoDB table would also require sending the sensor data to the cloud for inference, which would have the same drawbacks as option", + "references": "AWS Greengrass Machine Learning Inference - Amazon Web Services Machine learning components - AWS IoT Greengrass What is AWS Greengrass? | AWS IoT Core | Onica GitHub - aws-samples/aws-greengrass-ml-deployment-s ample AWS IoT Greengrass Architecture and Its Benefits | Quick Guide - XenonStack" + }, + { + "question": "A Machine Learning Specialist is designing a scalab le data storage solution for Amazon SageMaker. There is an existing TensorFlow-based model impleme nted as a train.py script that relies on static training data that is currently stored as TFRecords . Which method of providing training data to Amazon S ageMaker would meet the business requirements with the LEAST development overhead?", + "options": [ + "A. Use Amazon SageMaker script mode and use train.py unchanged. Point the Amazon", + "B. Use Amazon SageMaker script mode and use train.py unchanged. Put the TFRecord data into", + "C. Rewrite the train.py script to add a section that converts TFRecords to protobuf and ingests", + "D. Prepare the data in the format accepted by Amazon SageMaker. Use AWS Glue or AWS" + ], + "correct": "B. Use Amazon SageMaker script mode and use train.py unchanged. Put the TFRecord data into", + "explanation": "Amazon SageMaker script mode is a feature that allo ws users to use training scripts similar to those they would use outside SageMaker with SageMakers pr ebuilt containers for various frameworks such as TensorFlow. Script mode supports reading da ta from Amazon S3 buckets without requiring any changes to the training script. Therefore, opti on B is the best method of providing training data to Amazon SageMaker that would meet the business re quirements with the least development overhead. Option A is incorrect because using a local path of the data would not be scalable or reliable, as it would depend on the availability and capacity of th e local storage. Moreover, using a local path of the data would not leverage the benefits of Amazon S3, such as durability, security, and performance. Option C is incorrect because rewritin g the train.py script to convert TFRecords to protobuf would require additional development effor t and complexity, as well as introduce potential errors and inconsistencies in the data format. Opti on D is incorrect because preparing the data in the format accepted by Amazon SageMaker would also requ ire additional development effort and complexity, as well as involve using additional ser vices such as AWS Glue or AWS Lambda, which would increase the cost and maintenance of the solu tion.", + "references": "Bring your own model with Amazon SageMaker script m ode GitHub - aws-samples/amazon-sagemaker-script-mode Deep Dive on TensorFlow training with Amazon SageMa ker and Amazon S3 amazon-sagemaker-script-mode/generate_cifar10_tfrec ords.py at master" + }, + { + "question": "The chief editor for a product catalog wants the re search and development team to build a machine learning system that can be used to detect whether or not individuals in a collection of images are wearing the company's retail brand. The team has a set of training data. Which machine learning algorithm should the researc hers use that BEST meets their requirements?", + "options": [ + "A. Latent Dirichlet Allocation (LDA)", + "B. Recurrent neural network (RNN)", + "C. K-means", + "D. Convolutional neural network (CNN)" + ], + "correct": "D. Convolutional neural network (CNN)", + "explanation": "The problem of detecting whether or not individuals in a collection of images are wearing the companys retail brand is an example of image recogn ition, which is a type of machine learning task that identifies and classifies objects in an image. Convolutional neural networks (CNNs) are a type of machine learning algorithm that are well-suited for image recognition, as they can learn to extract features from images and handle variations in size, shape, color, and orientation of the objects. CNNs consist of multiple layers that perform convolution , pooling, and activation operations on the input images, resulting in a high-level representation th at can be used for classification or detection. Therefore, option D is the best choice for the mach ine learning algorithm that meets the requirements of the chief editor. Option A is incorrect because latent Dirichlet allo cation (LDA) is a type of machine learning algorith m that is used for topic modeling, which is a task th at discovers the hidden themes or topics in a collection of text documents. LDA is not suitable f or image recognition, as it does not preserve the spatial information of the pixels. Option B is inco rrect because recurrent neural networks (RNNs) are a type of machine learning algorithm that are used for sequential data, such as text, speech, or time series. RNNs can learn from the temporal dependenci es and patterns in the input data, and generate outputs that depend on the previous states. RNNs ar e not suitable for image recognition, as they do not capture the spatial dependencies and patterns i n the input images. Option C is incorrect because k-means is a type of machine learning algorithm tha t is used for clustering, which is a task that groups similar data points together based on their features. K-means is not suitable for image recognition, as it does not perform classification or detection of the objects in the images.", + "references": "Image Recognition Software - ML Image & Video Analy sis - Amazon \u00a6 Image classification and object detection using Ama zon Rekognition \u00a6 AWS Amazon Rekognition - Deep Learning Face and Ima ge Recognition \u00a6 GitHub - awslabs/aws-ai-solution-kit: Machine Learn ing APIs for common \u00a6 Meet iNaturalist, an AWS-powered nature app that he lps you identify \u00a6" + }, + { + "question": "A retail company is using Amazon Personalize to pro vide personalized product recommendations for its customers during a marketing campaign. The comp any sees a significant increase in sales of recommended items to existing customers immediately after deploying a new solution version, but these sales decrease a short time after deployment. Only historical data from before the marketing campaign is available for training. How should a data scientist adjust the solution?", + "options": [ + "A. Use the event tracker in Amazon Personalize to in clude real-time user interactions.", + "B. Add user metadata and use the HRNN-Metadata recip e in Amazon Personalize.", + "C. Implement a new solution using the built-in facto rization machines (FM) algorithm in", + "D. Add event type and event value fields to the inte ractions dataset in Amazon Personalize." + ], + "correct": "A. Use the event tracker in Amazon Personalize to in clude real-time user interactions.", + "explanation": "The best option is to use the event tracker in Amaz on Personalize to include real-time user interactions. This will allow the model to learn fr om the feedback of the customers during the marketing campaign and adjust the recommendations a ccordingly. The event tracker can capture click-through, add-to-cart, purchase, and other typ es of events that indicate the users preferences. By using the event tracker, the company can improve the relevance and freshness of the recommendations and avoid the decrease in sales. The other options are not as effective as using the event tracker. Adding user metadata and using the HRNN-Metadata recipe in Amazon Personalize can help capture the users attributes and preferences, but it will not reflect the changes in user behavior during the marketing campaign. Implementing a new solution using the built-in fact orization machines (FM) algorithm in Amazon SageMaker can also provide personalized recommendat ions, but it will require more time and effort to train and deploy the model. Adding event type an d event value fields to the interactions dataset in Amazon Personalize can help capture the importance and context of each interaction, but it will not update the model with the latest user feedback.", + "references": "Recording events - Amazon Personalize Using real-time events - Amazon Personalize" + }, + { + "question": "A machine learning (ML) specialist wants to secure calls to the Amazon SageMaker Service API. The specialist has configured Amazon VPC with a VPC int erface endpoint for the Amazon SageMaker Service API and is attempting to secure traffic fro m specific sets of instances and IAM users. The VPC is configured with a single public subnet. Which combination of steps should the ML specialist take to secure the traffic? (Choose two.)", + "options": [ + "A. Add a VPC endpoint policy to allow access to the IAM users.", + "B. Modify the users' IAM policy to allow access to A mazon SageMaker Service API calls only.", + "C. Modify the security group on the endpoint network interface to restrict access to the", + "D. Modify the ACL on the endpoint network interface to restrict access to the instances." + ], + "correct": "", + "explanation": "To secure calls to the Amazon SageMaker Service API , the ML specialist should take the following steps: Modify the security group on the endpoint network i nterface to restrict access to the instances. This will allow the ML specialist to control which insta nces in the VPC can communicate with the VPC interface endpoint for the Amazon SageMaker Service API. The security group can specify inbound and outbound rules based on the instance IDs, IP ad dresses, or CIDR blocks1. Add a SageMaker Runtime VPC endpoint interface to t he VPC. This will allow the ML specialist to invoke the SageMaker endpoints from within the VPC without using the public internet. The SageMaker Runtime VPC endpoint interface connects t he VPC directly to the SageMaker Runtime using AWS PrivateLink2. The other options are not as effective or necessary as the steps above. Adding a VPC endpoint policy to allow access to the IAM users is not required, a s the IAM users can already access the Amazon SageMaker Service API through the VPC interface end point. Modifying the users IAM policy to allow access to Amazon SageMaker Service API calls only i s not sufficient, as it does not prevent unauthorized instances from accessing the VPC inter face endpoint. Modifying the ACL on the endpoint network interface to restrict access to th e instances is not possible, as network ACLs are associated with subnets, not network interfaces3.", + "references": "Security groups for your VPC - Amazon Virtual Priva te Cloud Connect to SageMaker Within your VPC - Amazon SageM aker Network ACLs - Amazon Virtual Private Cloud" + }, + { + "question": "An e commerce company wants to launch a new cloud-b ased product recommendation feature for its web application. Due to data localization regul ations, any sensitive data must not leave its onpre mises data center, and the product recommendation model m ust be trained and tested using nonsensitive data only. Data transfer to the cloud must use IPsec. The web application is hosted on premises with a PostgreSQL database that contains a ll the dat", + "options": [ + "A. The company wants the data to be uploaded securel y to Amazon S3 each day for model retraining.", + "B. Create an AWS Glue job to connect to the PostgreS QL DB instance. Ingest tables without", + "C. Create an AWS Glue job to connect to the PostgreS QL DB instance. Ingest all data through an", + "D. Use AWS Database Migration Service (AWS DMS) with table mapping to select PostgreSQL" + ], + "correct": "C. Create an AWS Glue job to connect to the PostgreS QL DB instance. Ingest all data through an", + "explanation": "The best option is to use AWS Database Migration Se rvice (AWS DMS) with table mapping to select PostgreSQL tables with no sensitive data through an SSL connection. Replicate data directly into Amazon S3. This option meets the following requirem ents: It ensures that only nonsensitive data is transferr ed to the cloud by using table mapping to filter ou t the tables that contain sensitive data1. It uses IPsec to secure the data transfer by enabli ng SSL encryption for the AWS DMS endpoint2. It uploads the data to Amazon S3 each day for model retraining by using the ongoing replication feature of AWS DMS3. The other options are not as effective or feasible as the option above. Creating an AWS Glue job to connect to the PostgreSQL DB instance and ingest da ta through an AWS Site-to-Site VPN connection directly into Amazon S3 is possible, but it require s more steps and resources than using AWS DMS. Also, it does not specify how to filter out the sen sitive data from the tables. Creating an AWS Glue job to connect to the PostgreSQL DB instance and in gest all data through an AWS Site-to-Site VPN connection into Amazon S3 while removing sensitive data using a PySpark job is also possible, but it is more complex and error-prone than using AWS DMS. Also, it does not use IPsec as required. Using PostgreSQL logical replication to replicate all dat a to PostgreSQL in Amazon EC2 through AWS Direct Connect with a VPN connection, and then using AWS G lue to move data from Amazon EC2 to Amazon S3 is not feasible, because PostgreSQL logic al replication does not support replicating only a subset of data4. Also, it involves unnecessary data movement and additional costs.", + "references": "Table mapping - AWS Database Migration Service Using SSL to encrypt a connection to a DB instance - AWS Database Migration Service Ongoing replication - AWS Database Migration Servic e Logical replication - PostgreSQL" + }, + { + "question": "A logistics company needs a forecast model to predi ct next month's inventory requirements for a single item in 10 warehouses. A machine learning sp ecialist uses Amazon Forecast to develop a forecast model from 3 years of monthly dat", + "options": [ + "A. There is no missing data. The specialist selects the DeepAR+ algorithm to train a predictor. The", + "B. Set PerformAutoML to true.", + "C. Set ForecastHorizon to 4.", + "D. Set ForecastFrequency to W for weekly." + ], + "correct": "", + "explanation": "The MAPE of the predictor could be improved by maki ng the following changes to the CreatePredictor API call: Set PerformAutoML to true. This will allow Amazon F orecast to automatically evaluate different algorithms and choose the one that minimizes the ob jective function, which is the mean of the weighted losses over the forecast types. By default , these are the p10, p50, and p90 quantile losses1. This option can help find a better algorithm than D eepAR+ for the given data. Set PerformHPO to true. This will enable hyperparam eter optimization (HPO), which is the process of finding the optimal values for the algorithm-specif ic parameters that affect the quality of the forecasts. HPO can improve the accuracy of the pred ictor by tuning the hyperparameters based on the training data2. The other options are not likely to improve the MAP E of the predictor. Setting ForecastHorizon to 4 will reduce the number of time steps that the model predicts, which may not match the business requirement of predicting next months inventory. Se tting ForecastFrequency to W for weekly will change the granularity of the forecasts, which may not be appropriate for the monthly data. Setting FeaturizationMethodName to filling will not have an y effect, since there is no missing data in the dataset.", + "references": "CreatePredictor - Amazon Forecast HPOConfig - Amazon Forecast" + }, + { + "question": "a retail company. The company has provided a datase t of historic inventory demand for its products as a .csv file stored in an Amazon S3 bucket. The t able below shows a sample of the dataset. How should the data scientist transform the data?", + "options": [ + "A. Use ETL jobs in AWS Glue to separate the dataset into a target time series dataset and an", + "B. Use a Jupyter notebook in Amazon SageMaker to sep arate the dataset into a related time", + "C. Use AWS Batch jobs to separate the dataset into a target time series dataset, a related time", + "D. Use a Jupyter notebook in Amazon SageMaker to tra nsform the data into the optimized" + ], + "correct": "A. Use ETL jobs in AWS Glue to separate the dataset into a target time series dataset and an", + "explanation": "Amazon Forecast requires the input data to be in a specific format. The data scientist should use ETL jobs in AWS Glue to separate the dataset into a tar get time series dataset and an item metadata dataset. The target time series dataset should cont ain the timestamp, item_id, and demand columns, while the item metadata dataset should contain the item_id, category, and lead_time columns. Both datasets should be uploaded as .csv files to Amazon S3 . Reference: How Amazon Forecast Works - Amazon Forecast Choosing Datasets - Amazon Forecast", + "references": "" + }, + { + "question": "A machine learning specialist is running an Amazon SageMaker endpoint using the built-in object detection algorithm on a P3 instance for real-time predictions in a company's production application. When evaluating the model's resource utilization, t he specialist notices that the model is using only a fraction of the GPU. Which architecture changes would ensure that provis ioned resources are being utilized effectively?", + "options": [ + "A. Redeploy the model as a batch transform job on an M5 instance.", + "B. Redeploy the model on an M5 instance. Attach Amaz on Elastic Inference to the instance.", + "C. Redeploy the model on a P3dn instance.", + "D. Deploy the model onto an Amazon Elastic Container Service (Amazon ECS) cluster using a P3" + ], + "correct": "B. Redeploy the model on an M5 instance. Attach Amaz on Elastic Inference to the instance.", + "explanation": "The best way to ensure that provisioned resources a re being utilized effectively is to redeploy the model on an M5 instance and attach Amazon Elastic I nference to the instance. Amazon Elastic Inference allows you to attach low-cost GPU-powered acceleration to Amazon EC2 and Amazon SageMaker instances to reduce the cost of running d eep learning inference by up to 75%. By using Amazon Elastic Inference, you can choose the instan ce type that is best suited to the overall CPU and memory needs of your application, and then separate ly configure the amount of inference acceleration that you need with no code changes. Th is way, you can avoid wasting GPU resources and pay only for what you use. Option A is incorrect because a batch transform job is not suitable for real-time predictions. Batch transform is a high-performance and cost-effective feature for generating inferences using your trained models. Batch transform manages all of the compute resources required to get inferences. Batch transform is ideal for scenarios where youre working with large batches of data, dont need sub-second latency, or need to process data that is stored in Amazon S3. Option C is incorrect because redeploying the model on a P3dn instance would not improve the resource utilization. P3dn instances are designed f or distributed machine learning and high performance computing applications that need high n etwork throughput and packet rate performance. They are not optimized for inference w orkloads. Option D is incorrect because deploying the model o nto an Amazon ECS cluster using a P3 instance would not ensure that provisioned resources are bei ng utilized effectively. Amazon ECS is a fully managed container orchestration service that allows you to run and scale containerized applications on AWS. However, using Amazon ECS would not address the issue of underutilized GPU resources. In fact, it might introduce additional overhead and co mplexity in managing the cluster.", + "references": "Amazon Elastic Inference - Amazon SageMaker Batch Transform - Amazon SageMaker Amazon EC2 P3 Instances Amazon EC2 P3dn Instances Amazon Elastic Container Service" + }, + { + "question": "A data scientist uses an Amazon SageMaker notebook instance to conduct data exploration and analysis. This requires certain Python packages tha t are not natively available on Amazon SageMaker to be installed on the notebook instance. How can a machine learning specialist ensure that r equired packages are automatically available on the notebook instance for the data scientist to use ?", + "options": [ + "A. Install AWS Systems Manager Agent on the underlyi ng Amazon EC2 instance and use Systems", + "B. Create a Jupyter notebook file (.ipynb) with cell s containing the package installation", + "C. Use the conda package manager from within the Jup yter notebook console to apply the", + "D. Create an Amazon SageMaker lifecycle configuratio n with package installation commands" + ], + "correct": "D. Create an Amazon SageMaker lifecycle configuratio n with package installation commands", + "explanation": "The best way to ensure that required packages are a utomatically available on the notebook instance for the data scientist to use is to create an Amazo n SageMaker lifecycle configuration with package installation commands and assign the lifecycle conf iguration to the notebook instance. A lifecycle configuration is a shell script that runs when you create or start a notebook instance. You can use a lifecycle configuration to customize the notebook i nstance by installing libraries, changing environment variables, or downloading datasets. You can also use a lifecycle configuration to automate the installation of custom Python packages that are not natively available on Amazon SageMaker. Option A is incorrect because installing AWS System s Manager Agent on the underlying Amazon EC2 instance and using Systems Manager Automation to ex ecute the package installation commands is not a recommended way to customize the notebook ins tance. Systems Manager Automation is a feature that lets you safely automate common and re petitive IT operations and tasks across AWS resources. However, using Systems Manager Automatio n would require additional permissions and configurations, and it would not guarantee that the packages are installed before the notebook instance is ready to use. Option B is incorrect because creating a Jupyter no tebook file (.ipynb) with cells containing the package installation commands to execute and placin g the file under the /etc/init directory of each Amazon SageMaker notebook instance is not a valid w ay to customize the notebook instance. The /etc/init directory is used to store scripts that a re executed during the boot process of the operatin g system, not the Jupyter notebook application. Moreo ver, a Jupyter notebook file is not a shell script that can be executed by the operating system. Option C is incorrect because using the conda packa ge manager from within the Jupyter notebook console to apply the necessary conda packages to th e default kernel of the notebook is not an automatic way to customize the notebook instance. T his option would require the data scientist to manually run the conda commands every time they cre ate or start a new notebook instance. This would not be efficient or convenient for the data s cientist.", + "references": "Customize a notebook instance using a lifecycle con figuration script - Amazon SageMaker AWS Systems Manager Automation - AWS Systems Manage r Conda environments - Amazon SageMaker" + }, + { + "question": "A data scientist needs to identify fraudulent user accounts for a company's ecommerce platform. The company wants the ability to determine if a newly c reated account is associated with a previously known fraudulent user. The data scientist is using AWS Glue to cleanse the company's application logs during ingestion. Which strategy will allow the data scientist to ide ntify fraudulent accounts?", + "options": [ + "A. Execute the built-in FindDuplicates Amazon Athena query.", + "B. Create a FindMatches machine learning transform i n AWS Glue.", + "C. Create an AWS Glue crawler to infer duplicate acc ounts in the source data.", + "D. Search for duplicate accounts in the AWS Glue Dat a Catalog." + ], + "correct": "B. Create a FindMatches machine learning transform i n AWS Glue.", + "explanation": "The best strategy to identify fraudulent accounts i s to create a FindMatches machine learning transform in AWS Glue. The FindMatches transform en ables you to identify duplicate or matching records in your dataset, even when the records do n ot have a common unique identifier and no fields match exactly. This can help you improve fraud dete ction by finding accounts that are associated with a previously known fraudulent user. You can teach t he FindMatches transform your definition of a oeduplicate or a oematch through examples, and it w ill use machine learning to identify other potential duplicates or matches in your dataset. Yo u can then use the FindMatches transform in your AWS Glue ETL jobs to cleanse your data. Option A is incorrect because there is no built-in FindDuplicates Amazon Athena query. Amazon Athena is an interactive query service that makes i t easy to analyze data in Amazon S3 using standard SQL. However, Amazon Athena does not provide a pred efined query to find duplicate records in a dataset. You would have to write your own SQL query to perform this task, which might not be as effective or accurate as using the FindMatches tran sform. Option C is incorrect because creating an AWS Glue crawler to infer duplicate accounts in the source data is not a valid strategy. An AWS Glue crawler i s a program that connects to a data store, progresses through a prioritized list of classifier s to determine the schema for your data, and then creates metadata tables in the AWS Glue Data Catalo g. A crawler does not perform any data cleansing or record matching tasks. Option D is incorrect because searching for duplica te accounts in the AWS Glue Data Catalog is not a feasible strategy. The AWS Glue Data Catalog is a c entral repository to store structural and operational metadata for your data assets. The Data Catalog does not store the actual data, but rather the metadata that describes where the data i s located, how it is formatted, and what it contains. Therefore, you cannot search for duplicat e records in the Data Catalog.", + "references": "Record matching with AWS Lake Formation FindMatches - AWS Glue Amazon Athena \" Interactive SQL Queries for Data in Amazon S3 AWS Glue Crawlers - AWS Glue AWS Glue Data Catalog - AWS Glue" + }, + { + "question": "A Data Scientist is developing a machine learning m odel to classify whether a financial transaction is fraudulent. The labeled data available for training consists of 100,000 non-fraudulent observations and 1,000 fraudulent observations. The Data Scientist applies the XGBoost algorithm to the data, resulting in the following confusion matrix when the trained model is applied to a previ ously unseen validation dataset. The accuracy of the model is 99.1%, but the Data Scientist needs to reduce the number of false negatives. Which combination of steps should the Data Scientis t take to reduce the number of false negative predictions by the model? (Choose two.)", + "options": [ + "A. Change the XGBoost eval_metric parameter to optim ize based on Root Mean Square Error", + "B. Increase the XGBoost scale_pos_weight parameter t o adjust the balance of positive and", + "C. Increase the XGBoost max_depth parameter because the model is currently underfitting the", + "D. Change the XGBoost eval_metric parameter to optim ize based on Area Under the ROC Curve" + ], + "correct": "", + "explanation": "The Data Scientist should increase the XGBoost scal e_pos_weight parameter to adjust the balance of positive and negative weights and change the XGBoos t eval_metric parameter to optimize based on Area Under the ROC Curve (AUC). This will help redu ce the number of false negative predictions by the model. The scale_pos_weight parameter controls the balance of positive and negative weights in the XGBoost algorithm. It is useful for imbalanced clas sification problems, such as fraud detection, where the number of positive examples (fraudulent transac tions) is much smaller than the number of negative examples (non-fraudulent transactions). By increasing the scale_pos_weight parameter, the Data Scientist can assign more weight to the positi ve class and make the model more sensitive to detecting fraudulent transactions. The eval_metric parameter specifies the metric that is used to measure the performance of the model during training and validation. The default m etric for binary classification problems is the error rate, which is the fraction of incorrect pred ictions. However, the error rate is not a good metr ic for imbalanced classification problems, because it does not take into account the cost of different types of errors. For example, in fraud detection, a false negative (failing to detect a fraudulent transaction) is more costly than a false positive ( flagging a non-fraudulent transaction as fraudulent ). Therefore, the Data Scientist should use a metric t hat reflects the trade-off between the true positiv e rate (TPR) and the false positive rate (FPR), such as the Area Under the ROC Curve (AUC). The AUC is a measure of how well the model can distinguish bet ween the positive and negative classes, regardless of the classification threshold. A highe r AUC means that the model can achieve a higher TPR with a lower FPR, which is desirable for fraud detection.", + "references": "XGBoost Parameters - Amazon Machine Learning Using XGBoost with Amazon SageMaker - AWS Machine L earning Blog" + }, + { + "question": "A data scientist has developed a machine learning t ranslation model for English to Japanese by using Amazon SageMaker's built-in seq2seq algorithm with 500,000 aligned sentence pairs. While testing with sample sentences, the data scientist finds tha t the translation quality is reasonable for an example as short as five words. However, the qualit y becomes unacceptable if the sentence is 100 words long. Which action will resolve the problem?", + "options": [ + "A. Change preprocessing to use n-grams.", + "B. Add more nodes to the recurrent neural network (R NN) than the largest sentence's word", + "C. Adjust hyperparameters related to the attention m echanism.", + "D. Choose a different weight initialization type." + ], + "correct": "C. Adjust hyperparameters related to the attention m echanism.", + "explanation": "The data scientist should adjust hyperparameters re lated to the attention mechanism to resolve the problem. The attention mechanism is a technique tha t allows the decoder to focus on different parts of the input sequence when generating the output se quence. It helps the model cope with long input sequences and improve the translation quality. The Amazon SageMaker seq2seq algorithm supports different types of attention mechanisms, such as do t, general, concat, and mlp. The data scientist can use the hyperparameter attention_type to choose the type of attention mechanism. The data scientist can also use the hyperparameter attention _coverage_type to enable coverage, which is a mechanism that penalizes the model for attending to the same input positions repeatedly. By adjusting these hyperparameters, the data scientist can fine-tune the attention mechanism and reduce the number of false negative predictions by the model.", + "references": "Sequence-to-Sequence Algorithm - Amazon SageMaker Attention Mechanism - Sockeye Documentation" + }, + { + "question": "A financial company is trying to detect credit card fraud. The company observed that, on average, 2% of credit card transactions were fraudulent. A data scientist trained a classifier on a year's worth o f credit card transactions dat", + "options": [ + "A. The model needs to identify the fraudulent transa ctions (positives) from the regular ones", + "B. Specificity", + "C. False positive rate", + "D. Accuracy" + ], + "correct": "", + "explanation": "The data scientist should use the area under the pr ecision-recall curve and the true positive rate to optimize the model. These metrics are suitable for imbalanced classification problems, such as credit card fraud detection, where the positive class (fra udulent transactions) is much rarer than the negative class (non-fraudulent transactions). The area under the precision-recall curve (AUPRC) i s a measure of how well the model can identify the positive class among all the predicted positive s. Precision is the fraction of predicted positives that are actually positive, and recall is the fract ion of actual positives that are correctly predicte d. A higher AUPRC means that the model can achieve a hig her precision with a higher recall, which is desirable for fraud detection. The true positive rate (TPR) is another name for re call. It is also known as sensitivity or hit rate. It measures the proportion of actual positives that ar e correctly identified by the model. A higher TPR means that the model can capture more positives, wh ich is the companys goal.", + "references": "Metrics for Imbalanced Classification in Python - M achine Learning Mastery Precision-Recall - scikit-learn" + }, + { + "question": "A machine learning specialist is developing a proof of concept for government users whose primary concern is security. The specialist is using Amazon SageMaker to train a convolutional neural network (CNN) model for a photo classifier application. The specialist wants to protect the data so that it cannot be accessed and transferred to a remote host by malicious code accidentally installed on the training container. Which action will provide the MOST secure protectio n?", + "options": [ + "A. Remove Amazon S3 access permissions from the Sage Maker execution role.", + "B. Encrypt the weights of the CNN model.", + "C. Encrypt the training and validation dataset.", + "D. Enable network isolation for training jobs." + ], + "correct": "D. Enable network isolation for training jobs.", + "explanation": "The most secure action to protect the data from bei ng accessed and transferred to a remote host by malicious code accidentally installed on the traini ng container is to enable network isolation for training jobs. Network isolation is a feature that allows you to run training and inference containers in internet-free mode, which blocks any outbound ne twork calls from the containers, even to other AWS services such as Amazon S3. Additionally, no AW S credentials are made available to the container runtime environment. This way, you can pr event unauthorized access to your data and resources by malicious code or users. You can enabl e network isolation by setting the EnableNetworkIsolation parameter to True when you c all CreateTrainingJob, CreateHyperParameterTuningJob, or CreateModel.", + "references": "Run Training and Inference Containers in Internet-F ree Mode - Amazon SageMaker" + }, + { + "question": "A medical imaging company wants to train a computer vision model to detect areas of concern on patients' CT scans. The company has a large collect ion of unlabeled CT scans that are linked to each patient and stored in an Amazon S3 bucket. The scan s must be accessible to authorized users only. A machine learning engineer needs to build a labeling pipeline. Which set of steps should the engineer take to buil d the labeling pipeline with the LEAST effort?", + "options": [ + "A. Create a workforce with AWS Identity and Access M anagement (IAM). Build a labeling tool", + "B. Create an Amazon Mechanical Turk workforce and ma nifest file. Create a labeling job by", + "C. Create a private workforce and manifest file. Cre ate a labeling job by using the built-in", + "D. Create a workforce with Amazon Cognito. Build a l abeling web application with AWS Amplify." + ], + "correct": "C. Create a private workforce and manifest file. Cre ate a labeling job by using the built-in", + "explanation": "The engineer should create a private workforce and manifest file, and then create a labeling job by using the built-in bounding box task type in Amazon SageMaker Ground Truth. This will allow the engineer to build the labeling pipeline with the le ast effort. A private workforce is a group of workers that you manage and who have access to your labeling tasks. You can use a private workforce to label sen sitive data that requires confidentiality, such as medical images. You can create a private workforce by using Amazon Cognito and inviting workers by email. You can also use AWS Single Sign-On or your own authentication system to manage your private workforce. A manifest file is a JSON file that lists the Amazo n S3 locations of your input data. You can use a manifest file to specify the data objects that you want to label in your labeling job. You can create a manifest file by using the AWS CLI, the AWS SDK, or the Amazon SageMaker console. A labeling job is a process that sends your input d ata to workers for labeling. You can use the Amazon SageMaker console to create a labeling job and choo se from several built-in task types, such as image classification, text classification, semantic segmentation, and bounding box. A bounding box task type allows workers to draw boxes around objec ts in an image and assign labels to them. This is suitable for object detection tasks, such as identi fying areas of concern on CT scans.", + "references": "Create and Manage Workforces - Amazon SageMaker Use Input and Output Data - Amazon SageMaker Create a Labeling Job - Amazon SageMaker Bounding Box Task Type - Amazon SageMaker" + }, + { + "question": "A company is using Amazon Textract to extract textu al data from thousands of scanned text-heavy legal documents daily. The company uses this inform ation to process loan applications automatically. Some of the documents fail business validation and are returned to human reviewers, who investigate the errors. This activity increases the time to process the loan applications. What should the company do to reduce the processing time of loan applications?", + "options": [ + "A. Configure Amazon Textract to route low-confidence predictions to Amazon SageMaker", + "B. Use an Amazon Textract synchronous operation inst ead of an asynchronous operation.", + "C. Configure Amazon Textract to route low-confidence predictions to Amazon Augmented AI", + "D. Use Amazon Rekognition's feature to detect text i n an image to extract the data from" + ], + "correct": "", + "explanation": "The company should configure Amazon Textract to rou te low-confidence predictions to Amazon Augmented AI (Amazon A2I). Amazon A2I is a service that allows you to implement human review of machine learning (ML) predictions. It also comes in tegrated with some of the Artificial Intelligence (AI) services such as Amazon Textract. By using Ama zon A2I, the company can perform a manual review on those words that have low confidence scor es before performing a business validation. This will help reduce the processing time of loan applic ations by avoiding errors and rework. Option A is incorrect because Amazon SageMaker Grou nd Truth is not a suitable service for human review of Amazon Textract predictions. Amazon SageM aker Ground Truth is a service that helps you build highly accurate training datasets for machine learning. It allows you to label your own data or use a workforce of human labelers. However, it does not provide an easy way to integrate with Amazon Textract and route low-confidence prediction s for human review. Option B is incorrect because using an Amazon Textr act synchronous operation instead of an asynchronous operation will not reduce the processi ng time of loan applications. A synchronous operation is a request-response operation that retu rns the results immediately. An asynchronous operation is a start-and-check operation that retur ns a job identifier that you can use to check the status and results later. The choice of operation d epends on the size and complexity of the document, not on the confidence of the predictions. Option D is incorrect because using Amazon Rekognit ions feature to detect text in an image to extract the data from scanned images is not a bette r alternative than using Amazon Textract. Amazon Rekognition is a service that provides computer vis ion capabilities, such as face recognition, object detection, and scene analysis. It can also detect t ext in an image, but it does not provide the same level of accuracy and functionality as Amazon Textr act. Amazon Textract can not only detect text, but also extract data from tables and forms, and unders tand the layout and structure of the document.", + "references": "Amazon Augmented AI Amazon SageMaker Ground Truth Amazon Textract Operations Amazon Rekognition" + }, + { + "question": "A company ingests machine learning (ML) data from w eb advertising clicks into an Amazon S3 data lake. Click data is added to an Amazon Kinesis data stream by using the Kinesis Producer Library (KPL). The data is loaded into the S3 data lake fro m the data stream by using an Amazon Kinesis Data Firehose delivery stream. As the data volume increa ses, an ML specialist notices that the rate of data ingested into Amazon S3 is relatively constant. The re also is an increasing backlog of data for Kinesi s Data Streams and Kinesis Data Firehose to ingest. Which next step is MOST likely to improve the data ingestion rate into Amazon S3?", + "options": [ + "A. Increase the number of S3 prefixes for the delive ry stream to write to.", + "B. Decrease the retention period for the data stream .", + "C. Increase the number of shards for the data stream .", + "D. Add more consumers using the Kinesis Client Libra ry (KCL)." + ], + "correct": "C. Increase the number of shards for the data stream .", + "explanation": "The data ingestion rate into Amazon S3 can be impro ved by increasing the number of shards for the data stream. A shard is the base throughput unit of a Kinesis data stream. One shard provides 1 MB/second data input and 2 MB/second data output. I ncreasing the number of shards increases the data ingestion capacity of the stream. This can hel p reduce the backlog of data for Kinesis Data Streams and Kinesis Data Firehose to ingest.", + "references": "Shard - Amazon Kinesis Data Streams Scaling Amazon Kinesis Data Streams with AWS CloudF ormation - AWS Big Data Blog" + }, + { + "question": "A data scientist must build a custom recommendation model in Amazon SageMaker for an online retail company. Due to the nature of the company's products, customers buy only 4-5 products every 5-10 years. So, the company relies on a steady stre am of new customers. When a new customer signs up, the company collects data on the customer's pre ferences. Below is a sample of the data available to the data scientist. How should the data scientist split the dataset int o a training and test set for this use case?", + "options": [ + "A. Shuffle all interaction dat", + "B. Split off the last 10% of the interaction data fo r the test set.", + "C. Identify the most recent 10% of interactions for each user. Split off these interactions for the", + "D. Identify the 10% of users with the least interact ion data. Split off all interaction data from" + ], + "correct": "D. Identify the 10% of users with the least interact ion data. Split off all interaction data from", + "explanation": "The best way to split the dataset into a training a nd test set for this use case is to randomly select 10% of the users and split off all interaction data from these users for the test set. This is because the company relies on a steady stream of new customers, so the test set should reflect the behavior of new customers who have not been seen by the model b efore. The other options are not suitable because they either mix old and new customers in th e test set (A and B), or they bias the test set towards users with less interaction data \u00c2\u00a9. Refere nce: Amazon SageMaker Developer Guide: Train and Test Da tasets Amazon Personalize Developer Guide: Preparing and I mporting Data", + "references": "" + }, + { + "question": "A financial services company wants to adopt Amazon SageMaker as its default data science environment. The company's data scientists run mach ine learning (ML) models on confidential financial dat", + "options": [ + "A. The company is worried about data egress and want s an ML engineer to secure the environment.", + "B. Connect to SageMaker by using a VPC interface end point powered by AWS PrivateLink.", + "C. Use SCPs to restrict access to SageMaker.", + "D. Disable root access on the SageMaker notebook insta nces. E. Enable network isolation for training jobs and mode ls." + ], + "correct": "", + "explanation": "Explanation/Reference: To control data egress from SageMaker, the ML engin eer can use the following mechanisms: Connect to SageMaker by using a VPC interface endpo int powered by AWS PrivateLink. This allows the ML engineer to access SageMaker services and re sources without exposing the traffic to the public internet. This reduces the risk of data leak age and unauthorized access1 Enable network isolation for training jobs and mode ls. This prevents the training jobs and models from accessing the internet or other AWS services. This ensures that the data used for training and inference is not exposed to external sources2 Protect data with encryption at rest and in transit . Use AWS Key Management Service (AWS KMS) to manage encryption keys. This enables the ML enginee r to encrypt the data stored in Amazon S3 buckets, SageMaker notebook instances, and SageMake r endpoints. It also allows the ML engineer to encrypt the data in transit between SageMaker and o ther AWS services. This helps protect the data from unauthorized access and tampering3 The other options are not effective in controlling data egress from SageMaker: Use SCPs to restrict access to SageMaker. SCPs are used to define the maximum permissions for an organization or organizational unit (OU) in AWS Org anizations. They do not control the data egress from SageMaker, but rather the access to SageMaker itself4 Disable root access on the SageMaker notebook insta nces. This prevents the users from installing additional packages or libraries on the notebook in stances. It does not prevent the data from being transferred out of the notebook instances. Restrict notebook presigned URLs to specific IPs us ed by the company. This limits the access to the notebook instances from certain IP addresses. It do es not prevent the data from being transferred out of the notebook instances.", + "references": "1: Amazon SageMaker Interface VPC Endpoints (AWS Pr ivateLink) - Amazon SageMaker 2: Network Isolation - Amazon SageMaker 3: Encrypt Data at Rest and in Transit - Amazon Sag eMaker 4: Using Service Control Policies - AWS Organizatio ns : Disable Root Access - Amazon SageMaker : Create a Presigned Notebook Instance URL - Amazon SageMaker" + }, + { + "question": "A company needs to quickly make sense of a large am ount of data and gain insight from it. The data is in different formats, the schemas change frequen tly, and new data sources are added regularly. The company wants to use AWS services to explore mu ltiple data sources, suggest schemas, and enrich and transform the dat", + "options": [ + "A. The solution should require the least possible co ding effort for the data flows and the least", + "B. Amazon EMR for data discovery, enrichment, and tr ansformation", + "C. Amazon Kinesis Data Analytics for data ingestion", + "D. AWS Glue for data discovery, enrichment, and tran sformation" + ], + "correct": "C. Amazon Kinesis Data Analytics for data ingestion", + "explanation": "The best combination of AWS services to meet the re quirements of data discovery, enrichment, transformation, querying, analysis, and reporting w ith the least coding and infrastructure management is AWS Glue, Amazon Athena, and Amazon Q uickSight. These services are: AWS Glue for data discovery, enrichment, and transf ormation. AWS Glue is a serverless data integration service that automatically crawls, cata logs, and prepares data from various sources and formats. It also provides a visual interface called AWS Glue DataBrew that allows users to apply over 250 transformations to clean, normalize, and enrich data without writing code1 Amazon Athena for querying and analyzing the result s in Amazon S3 using standard SQL. Amazon Athena is a serverless interactive query service th at allows users to analyze data in Amazon S3 using standard SQL. It supports a variety of data formats , such as CSV, JSON, ORC, Parquet, and Avro. It als o integrates with AWS Glue Data Catalog to provide a unified view of the data sources and schemas2 Amazon QuickSight for reporting and getting insight s. Amazon QuickSight is a serverless business intelligence service that allows users to create an d share interactive dashboards and reports. It also provides ML-powered features, such as anomaly detec tion, forecasting, and natural language queries, to help users discover hidden insights fro m their data3 The other options are not suitable because they eit her require more coding effort, more infrastructure management, or do not support the de sired use cases. For example: Option A uses Amazon EMR for data discovery, enrich ment, and transformation. Amazon EMR is a managed cluster platform that runs Apache Spark, Ap ache Hive, and other open-source frameworks for big data processing. It requires users to write code in languages such as Python, Scala, or SQL to perform data integration tasks. It also requires us ers to provision, configure, and scale the clusters according to their needs4 Option B uses Amazon Kinesis Data Analytics for dat a ingestion. Amazon Kinesis Data Analytics is a service that allows users to process streaming data in real time using SQL or Apache Flink. It is not suitable for data discovery, enrichment, and transf ormation, which are typically batch-oriented tasks. It also requires users to write code to defi ne the data processing logic and the output destination5 Option D uses AWS Data Pipeline for data transfer a nd AWS Step Functions for orchestrating AWS Lambda jobs for data discovery, enrichment, and tra nsformation. AWS Data Pipeline is a service that helps users move data between AWS services and on-p remises data sources. AWS Step Functions is a service that helps users coordinate multiple AWS se rvices into workflows. AWS Lambda is a service that lets users run code without provisioning or ma naging servers. These services require users to write code to define the data sources, destinations , transformations, and workflows. They also require users to manage the scalability, performanc e, and reliability of the data pipelines.", + "references": "1: AWS Glue - Data Integration Service - Amazon Web Services 2: Amazon Athena \" Interactive SQL Query Service - AWS 3: Amazon QuickSight - Business Intelligence Servic e - AWS 4: Amazon EMR - Amazon Web Services 5: Amazon Kinesis Data Analytics - Amazon Web Servi ces : AWS Data Pipeline - Amazon Web Services : AWS Step Functions - Amazon Web Services : AWS Lambda - Amazon Web Services" + }, + { + "question": "A company is converting a large number of unstructu red paper receipts into images. The company wants to create a model based on natural language p rocessing (NLP) to find relevant entities such as date, location, and notes, as well as some custom e ntities such as receipt numbers. The company is using optical character recognition (OCR) to extract text for data labeling. However, documents are in different structures and formats, and the company is facing challenges with setting up the manual workflows for each document type. Add itionally, the company trained a named entity recognition (NER) model for custom entity detection using a small sample size. This model has a very low confidence score and will require retraining wi th a large dataset. Which solution for text extraction and entity detec tion will require the LEAST amount of effort?", + "options": [ + "A. Extract text from receipt images by using Amazon Textract. Use the Amazon SageMaker", + "B. Extract text from receipt images by using a deep learning OCR model from the AWS Marketplace.", + "C. Extract text from receipt images by using Amazon Textract. Use Amazon Comprehend for entity", + "D. Extract text from receipt images by using a deep learning OCR model from the AWS Marketplace." + ], + "correct": "C. Extract text from receipt images by using Amazon Textract. Use Amazon Comprehend for entity", + "explanation": "The best solution for text extraction and entity de tection with the least amount of effort is to use Amazon Textract and Amazon Comprehend. These servic es are: Amazon Textract for text extraction from receipt im ages. Amazon Textract is a machine learning service that can automatically extract text and dat a from scanned documents. It can handle different structures and formats of documents, such as PDF, T IFF, PNG, and JPEG, without any preprocessing steps. It can also extract key-value pairs and tabl es from documents1 Amazon Comprehend for entity detection and custom e ntity detection. Amazon Comprehend is a natural language processing service that can identi fy entities, such as dates, locations, and notes, from unstructured text. It can also detect custom e ntities, such as receipt numbers, by using a custom entity recognizer that can be trained with a small amount of labeled data2 The other options are not suitable because they eit her require more effort for text extraction, entity detection, or custom entity detection. For example: Option A uses the Amazon SageMaker BlazingText algo rithm to train on the text for entities and custom entities. BlazingText is a supervised learni ng algorithm that can perform text classification and word2vec. It requires users to provide a large amount of labeled data, preprocess the data into a specific format, and tune the hyperparameters of th e model3 Option B uses a deep learning OCR model from the AW S Marketplace and a NER deep learning model for text extraction and entity detection. The se models are pre-trained and may not be suitable for the specific use case of receipt processing. Th ey also require users to deploy and manage the models on Amazon SageMaker or Amazon EC2 instances4 Option D uses a deep learning OCR model from the AW S Marketplace for text extraction. This model has the same drawbacks as option B. It also require s users to integrate the model output with Amazon Comprehend for entity detection and custom e ntity detection.", + "references": "1: Amazon Textract \" Extract text and data from doc uments 2: Amazon Comprehend \" Natural Language Processing (NLP) and Machine Learning (ML) 3: BlazingText - Amazon SageMaker 4: AWS Marketplace: OCR" + }, + { + "question": "A company is building a predictive maintenance mode l based on machine learning (ML). The data is stored in a fully private Amazon S3 bucket that is encrypted at rest with AWS Key Management Service (AWS KMS) CMKs. An ML specialist must run d ata preprocessing by using an Amazon SageMaker Processing job that is triggered from cod e in an Amazon SageMaker notebook. The job should read data from Amazon S3, process it, and up load it back to the same S3 bucket. The preprocessing code is stored in a container image i n Amazon Elastic Container Registry (Amazon ECR). The ML specialist needs to grant permissions to ensure a smooth data preprocessing workflow. Which set of actions should the ML specialist take to meet these requirements?", + "options": [ + "A. Create an IAM role that has permissions to create Amazon SageMaker Processing jobs, S3 read", + "B. Create an IAM role that has permissions to create Amazon SageMaker Processing jobs. Attach the", + "C. Create an IAM role that has permissions to create Amazon SageMaker Processing jobs and to", + "D. Create an IAM role that has permissions to create Amazon SageMaker Processing jobs. Attach the" + ], + "correct": "B. Create an IAM role that has permissions to create Amazon SageMaker Processing jobs. Attach the", + "explanation": "The correct solution for granting permissions for d ata preprocessing is to use the following steps: Create an IAM role that has permissions to create A mazon SageMaker Processing jobs. Attach the role to the SageMaker notebook instance. This role allows the ML specialist to run Processing jobs from the notebook code1 Create an Amazon SageMaker Processing job with an I AM role that has read and write permissions to the relevant S3 bucket, and appropriate KMS and ECR permissions. This role allows the Processing job to access the data in the encrypted S3 bucket, decrypt it with the KMS CMK, and pull the container image from ECR23 The other options are incorrect because they either miss some permissions or use unnecessary steps. For example: Option A uses a single IAM role for both the notebo ok instance and the Processing job. This role may have more permissions than necessary for the notebo ok instance, which violates the principle of least privilege4 Option C sets up both an S3 endpoint and a KMS endp oint in the default VPC. These endpoints are not required for the Processing job to access the d ata in the encrypted S3 bucket. They are only needed if the Processing job runs in network isolat ion mode, which is not specified in the question. Option D uses the access key and secret key of the IAM user with appropriate KMS and ECR permissions. This is not a secure way to pass crede ntials to the Processing job. It also requires the ML specialist to manage the IAM user and the keys.", + "references": "1: Create an Amazon SageMaker Notebook Instance - A mazon SageMaker 2: Create a Processing Job - Amazon SageMaker 3: Use AWS KMS\"Managed Encryption Keys - Amazon Sim ple Storage Service 4: IAM Best Practices - AWS Identity and Access Man agement : Network Isolation - Amazon SageMaker : Understanding and Getting Your Security Credentia ls - AWS General Reference" + }, + { + "question": "A data scientist has been running an Amazon SageMak er notebook instance for a few weeks. During this time, a new version of Jupyter Notebook was re leased along with additional software updates. The security team mandates that all running SageMak er notebook instances use the latest security and software updates provided by SageMaker. How can the data scientist meet these requirements?A. Call the CreateNotebookInstanceLifecycleConfig AP I operation", + "options": [ + "B. Create a new SageMaker notebook instance and moun t the Amazon Elastic Block Store (Amazon", + "C. Stop and then restart the SageMaker notebook inst ance", + "D. Call the UpdateNotebookInstanceLifecycleConfig AP I operation" + ], + "correct": "C. Stop and then restart the SageMaker notebook inst ance", + "explanation": "The correct solution for updating the software on a SageMaker notebook instance is to stop and then restart the notebook instance. This will automatica lly apply the latest security and software updates provided by SageMaker1 The other options are incorrect because they either do not update the software or require unnecessary steps. For example: Option A calls the CreateNotebookInstanceLifecycleC onfig API operation. This operation creates a lifecycle configuration, which is a set of shell sc ripts that run when a notebook instance is created or started. A lifecycle configuration can be used to c ustomize the notebook instance, such as installing additional libraries or packages. However, it does not update the software on the notebook instance2 Option B creates a new SageMaker notebook instance and mounts the Amazon Elastic Block Store (Amazon EBS) volume from the original instance. Thi s option will create a new notebook instance with the latest software, but it will also incur ad ditional costs and require manual steps to transfer the data and settings from the original instance3 Option D calls the UpdateNotebookInstanceLifecycleC onfig API operation. This operation updates an existing lifecycle configuration. As explained in o ption A, a lifecycle configuration does not update the software on the notebook instance4", + "references": "1: Amazon SageMaker Notebook Instances - Amazon Sag eMaker 2: CreateNotebookInstanceLifecycleConfig - Amazon S ageMaker 3: Create a Notebook Instance - Amazon SageMaker 4: UpdateNotebookInstanceLifecycleConfig - Amazon S ageMaker" + }, + { + "question": "A library is developing an automatic book-borrowing system that uses Amazon Rekognition. Images of library members faces are stored in an Amazon S3 bucket. When members borrow books, the Amazon Rekognition CompareFaces API operation compa res real faces against the stored faces in Amazon S3. The library needs to improve security by making sur e that images are encrypted at rest. Also, when the images are used with Amazon Rekognition. they n eed to be encrypted in transit. The library also must ensure that the images are not used to improve Amazon Rekognition as a service. How should a machine learning specialist architect the solution to satisfy these requirements?", + "options": [ + "A. Enable server-side encryption on the S3 bucket. S ubmit an AWS Support ticket to opt out of", + "B. Switch to using an Amazon Rekognition collection to store the images. Use the IndexFaces and", + "C. Switch to using the AWS GovCloud (US) Region for Amazon S3 to store images and for Amazon", + "D. Enable client-side encryption on the S3 bucket. S et up a VPN connection and only call the Amazon" + ], + "correct": "A. Enable server-side encryption on the S3 bucket. S ubmit an AWS Support ticket to opt out of", + "explanation": "The best solution for encrypting images at rest and in transit, and opting out of data usage for servi ce improvement, is to use the following steps: Enable server-side encryption on the S3 bucket. Thi s will encrypt the images stored in the bucket using AWS Key Management Service (AWS KMS) customer master keys (CMKs). This will protect the data at rest from unauthorized access1 Submit an AWS Support ticket to opt out of allowing images to be used for improving the service, and follow the process provided by AWS Support. Thi s will prevent AWS from storing or using the images processed by Amazon Rekognition for service development or enhancement purposes. This will protect the data privacy and ownership2 Use HTTPS to call the Amazon Rekognition CompareFac es API operation. This will encrypt the data in transit between the client and the server using SSL /TLS protocols. This will protect the data from interception or tampering3 The other options are incorrect because they either do not encrypt the images at rest or in transit, o r do not opt out of data usage for service improvemen t. For example: Option B switches to using an Amazon Rekognition co llection to store the images. A collection is a container for storing face vectors that are calcula ted by Amazon Rekognition. It does not encrypt the images at rest or in transit, and it does not opt o ut of data usage for service improvement. It also requires changing the API operations from CompareFa ces to IndexFaces and SearchFacesByImage, which may not have the same functionality or perfor mance4 Option C switches to using the AWS GovCloud (US) Re gion for Amazon S3 and Amazon Rekognition. The AWS GovCloud (US) Region is an isolated AWS Reg ion designed to host sensitive data and regulated workloads in the cloud. It does not autom atically encrypt the images at rest or in transit, and it does not opt out of data usage for service i mprovement. It also requires migrating the data and the application to a different Region, which ma y incur additional costs and complexity5 Option D enables client-side encryption on the S3 b ucket. This means that the client is responsible for encrypting and decrypting the images before upl oading or downloading them from the bucket. This adds extra overhead and complexity to the clie nt application, and it does not encrypt the data in transit when calling the Amazon Rekognition API. It also does not opt out of data usage for service improvement.", + "references": "1: Protecting Data Using Server-Side Encryption wit h AWS KMS\"Managed Keys (SSE-KMS) - Amazon Simple Storage Service 2: Opting Out of Content Storage and Use for Servic e Improvements - Amazon Rekognition 3: HTTPS - Wikipedia 4: Working with Stored Faces - Amazon Rekognition 5: AWS GovCloud (US) - Amazon Web Services : Protecting Data Using Client-Side Encryption - Am azon Simple Storage Service" + }, + { + "question": "A company is building a line-counting application f or use in a quick-service restaurant. The company wants to use video cameras pointed at the line of c ustomers at a given register to measure how many people are in line and deliver notifications t o managers if the line grows too long. The restaurant locations have limited bandwidth for con nections to external services and cannot accommodate multiple video streams without impactin g other operations. Which solution should a machine learning specialist implement to meet these requirements?", + "options": [ + "A. Install cameras compatible with Amazon Kinesis Vi deo Streams to stream the data to AWS over", + "B. Deploy AWS DeepLens cameras in the restaurant to capture video. Enable Amazon Rekognition on", + "C. Build a custom model in Amazon SageMaker to recog nize the number of people in an image. Install cameras compatible with Amazon Kinesis Vide o Streams in the restaurant. Write an AWS", + "D. Build a custom model in Amazon SageMaker to recog nize the number of people in an image.", + "A. Decrease the cooldown period for the scale-in act ivity. Increase the configured maximum capacity", + "B. Replace the current endpoint with a multi-model e ndpoint using SageMaker.", + "C. Set up Amazon API Gateway and AWS Lambda to trigg er the SageMaker inference endpoint.", + "D. Increase the cooldown period for the scale-out ac tivity." + ], + "correct": "D. Increase the cooldown period for the scale-out ac tivity.", + "explanation": ": The correct solution for changing the scaling beh avior of the SageMaker instances is to increase the cooldown period for the scale-out activity. The coo ldown period is the amount of time, in seconds, after a scaling activity completes before another s caling activity can start. By increasing the cooldo wn period for the scale-out activity, the ML team can ensure that the new instances are ready before launching additional instances. This will prevent o ver-scaling and reduce costs1 The other options are incorrect because they either do not solve the issue or require unnecessary steps. For example: Option A decreases the cooldown period for the scal e-in activity and increases the configured maximum capacity of instances. This option does not address the issue of launching additional instances before the new instances are ready. It ma y also cause under-scaling and performance degradation. Option B replaces the current endpoint with a multi -model endpoint using SageMaker. A multimodel endpoint is an endpoint that can host multiple mode ls using a single endpoint. It does not affect the scaling behavior of the SageMaker instan ces. It also requires creating a new endpoint and updating the application code to use it2 Option C sets up Amazon API Gateway and AWS Lambda to trigger the SageMaker inference endpoint. Amazon API Gateway is a service that allo ws users to create, publish, maintain, monitor, and secure APIs. AWS Lambda is a service that lets users run code without provisioning or managing servers. These services do not affect the scaling b ehavior of the SageMaker instances. They also require creating and configuring additional resourc es and services34", + "references": "1: Automatic Scaling - Amazon SageMaker 2: Create a Multi-Model Endpoint - Amazon SageMaker 3: Amazon API Gateway - Amazon Web Services 4: AWS Lambda - Amazon Web Services" + }, + { + "question": "A telecommunications company is developing a mobile app for its customers. The company is using an Amazon SageMaker hosted endpoint for machine lea rning model inferences. Developers want to introduce a new version of the m odel for a limited number of users who subscribed to a preview feature of the app. After t he new version of the model is tested as a preview, developers will evaluate its accuracy. If a new ver sion of the model has better accuracy, developers need to be able to gradually release the new versio n for all users over a fixed period of time. How can the company implement the testing model wit h the LEAST amount of operational overhead?", + "options": [ + "A. Update the ProductionVariant data type with the n ew version of the model by using the", + "B. Configure two SageMaker hosted endpoints that ser ve the different versions of the model. Create", + "C. Update the DesiredWeightsAndCapacity data type wi th the new version of the model by using the", + "D. Configure two SageMaker hosted endpoints that ser ve the different versions of the model. Create" + ], + "correct": "C. Update the DesiredWeightsAndCapacity data type wi th the new version of the model by using the", + "explanation": "The best solution for implementing the testing mode l with the least amount of operational overhead is to use the following steps: Update the DesiredWeightsAndCapacity data type with the new version of the model by using the UpdateEndpointWeightsAndCapacities operation with t he DesiredWeight parameter set to 0. This operation allows the developers to update the varia nt weights and capacities of an existing SageMaker endpoint without deleting and recreating the endpoint. Setting the DesiredWeight parameter to 0 means that the new version of the mo del will not receive any traffic initially1 Specify the TargetVariant parameter for InvokeEndpo int calls for users who subscribed to the preview feature. This parameter allows the developers to ov erride the variant weights and direct a request to a specific variant. This way, the developers can te st the new version of the model for a limited number of users who opted in for the preview featur e2 When the new version of the model is ready for rele ase, gradually increase DesiredWeight until all users have the updated version. This operation allo ws the developers to perform a gradual rollout of the new version of the model and monitor its perfor mance and accuracy. The developers can adjust the variant weights and capacities as needed until the new version of the model serves all the traffic1 The other options are incorrect because they either require more operational overhead or do not support the desired use cases. For example: Option A uses the CreateEndpointConfig operation wi th the InitialVariantWeight parameter set to 0. This operation creates a new endpoint configuration , which requires deleting and recreating the endpoint to apply the changes. This adds extra over head and downtime for the endpoint. It also does not support the gradual rollout of the new version of the model3 Option B uses two SageMaker hosted endpoints that s erve the different versions of the model and an Application Load Balancer (ALB) to route traffic to both endpoints based on the TargetVariant query string parameter. This option requires creating and managing additional resources and services, such as the second endpoint and the ALB. It also require s changing the app code to send the query string parameter for the preview feature4 Option D uses the access key and secret key of the IAM user with appropriate KMS and ECR permissions. This is not a secure way to pass crede ntials to the Processing job. It also requires the ML specialist to manage the IAM user and the keys.", + "references": "1: UpdateEndpointWeightsAndCapacities - Amazon Sage Maker 2: InvokeEndpoint - Amazon SageMaker 3: CreateEndpointConfig - Amazon SageMaker 4: Application Load Balancer - Elastic Load Balanci ng" + }, + { + "question": "A company offers an online shopping service to its customers. The company wants to enhance the sites security by requesting additional information when customers access the site from locations that are different from their normal location. The company wants to update the process to call a machine learning (ML) model to determine when addit ional information should be requested. The company has several terabytes of data from its existing ecommerce web servers containing the source IP addresses for each request made to the we b server. For authenticated requests, the records also contain the login name of the requesting user. Which approach should an ML specialist take to impl ement the new security feature in the web application?", + "options": [ + "A. Use Amazon SageMaker Ground Truth to label each r ecord as either a successful or failed access", + "B. Use Amazon SageMaker to train a model using the I P Insights algorithm. Schedule updates and", + "C. Use Amazon SageMaker Ground Truth to label each r ecord as either a successful or failed access", + "D. Use Amazon SageMaker to train a model using the O bject2Vec algorithm. Schedule updates and" + ], + "correct": "B. Use Amazon SageMaker to train a model using the I P Insights algorithm. Schedule updates and", + "explanation": "The IP Insights algorithm is designed to capture as sociations between entities and IP addresses, and can be used to identify anomalous IP usage patterns . The algorithm can learn from historical data that contains pairs of entities and IP addresses, a nd can return a score that indicates how likely the pair is to occur. The company can use this algorith m to train a model that can detect when a customer is accessing the site from a different loc ation than usual, and request additional information accordingly. The company can also sched ule updates and retraining of the model using new log data nightly to keep the model up to date w ith the latest IP usage patterns. The other options are not suitable for this use cas e because: Option A: The factorization machines (FM) algorithm is a general-purpose supervised learning algorithm that can be used for both classification and regression tasks. However, it is not optimized for capturing associations between entities and IP addresses, and would require labeling each record as either a successful or failed access attempt, wh ich is a costly and time-consuming process. Option C: The IP Insights algorithm is a good choic e for this use case, but it does not require labeli ng each record as either a successful or failed access attempt. The algorithm is unsupervised and can learn from the historical data without labels. Labe ling the data would be unnecessary and wasteful. Option D: The Object2Vec algorithm is a general-pur pose neural embedding algorithm that can learn low-dimensional dense embeddings of high-dimensiona l objects. However, it is not designed to capture associations between entities and IP addres ses, and would require a different input format than the one provided by the company. The Object2Ve c algorithm expects pairs of objects and their relationship labels or scores as inputs, while the company has data containing the source IP addresses and the login names of the requesting use rs.", + "references": "IP Insights - Amazon SageMaker Factorization Machines Algorithm - Amazon SageMaker Object2Vec Algorithm - Amazon SageMaker" + }, + { + "question": "A retail company wants to combine its customer orde rs with the product description data from its product catalog. The structure and format of the re cords in each dataset is different. A data analyst tried to use a spreadsheet to combine the datasets, but the effort resulted in duplicate records and records that were not properly combined. The compan y needs a solution that it can use to combine similar records from the two datasets and remove an y duplicates. Which solution will meet these requirements?", + "options": [ + "A. Use an AWS Lambda function to process the data. U se two arrays to compare equal strings in the", + "B. Create AWS Glue crawlers for reading and populati ng the AWS Glue Data Catalog. Call the AWS", + "C. Create AWS Glue crawlers for reading and populati ng the AWS Glue Data Catalog. Use the", + "D. Create an AWS Lake Formation custom transform. Ru n a transformation for matching products" + ], + "correct": "C. Create AWS Glue crawlers for reading and populati ng the AWS Glue Data Catalog. Use the", + "explanation": "The FindMatches transform is a machine learning tra nsform that can identify and match similar records from different datasets, even when the reco rds do not have a common unique identifier or exact field values. The FindMatches transform can a lso remove duplicate records from a single dataset. The FindMatches transform can be used with AWS Glue crawlers and jobs to process the data from various sources and store it in a data la ke. The FindMatches transform can be created and managed using the AWS Glue console, API, or AWS Glu e Studio. The other options are not suitable for this use cas e because: Option A: Using an AWS Lambda function to process t he data and compare equal strings in the fields from the two datasets is not an efficient or scalab le solution. It would require writing custom code and handling the data loading and cleansing logic. It would also not account for variations or inconsistencies in the field values, such as spelli ng errors, abbreviations, or missing data. Option B: The AWS Glue SearchTables API operation i s used to search for tables in the AWS Glue Data Catalog based on a set of criteria. It is not a mac hine learning transform that can match records across different datasets or remove duplicates. It would also require writing custom code to invoke the API and process the results. Option D: AWS Lake Formation does not provide a cus tom transform feature. It provides predefined blueprints for common data ingestion scenarios, suc h as database snapshot, incremental database, and log file. These blueprints do not support match ing records across different datasets or removing duplicates.", + "references": "" + }, + { + "question": "A company provisions Amazon SageMaker notebook inst ances for its data science team and creates Amazon VPC interface endpoints to ensure communicat ion between the VPC and the notebook instances. All connections to the Amazon SageMaker API are contained entirely and securely using the AWS network. However, the data science team rea lizes that individuals outside the VPC can still connect to the notebook instances across the intern et. Which set of actions should the data science team t ake to fix the issue?", + "options": [ + "A. Modify the notebook instances' security group to allow traffic only from the CIDR ranges of the", + "B. Create an IAM policy that allows the sagemaker:Cr eatePresignedNotebooklnstanceUrl and", + "C. Add a NAT gateway to the VPC. Convert all of the subnets where the Amazon SageMaker notebook", + "D. Change the network ACL of the subnet the notebook is hosted in to restrict access to anyone outside the VPC." + ], + "correct": "A. Modify the notebook instances' security group to allow traffic only from the CIDR ranges of the", + "explanation": "The issue is that the notebook instances security g roup allows inbound traffic from any source IP address, which means that anyone with the authorize d URL can access the notebook instances over the internet. To fix this issue, the data science t eam should modify the security group to allow traff ic only from the CIDR ranges of the VPC, which are the IP addresses assigned to the resources within the VPC. This way, only the VPC interface endpoints and the resources within the VPC can communicate with the notebook instances. The data s cience team should apply this security group to all of the notebook instances VPC interfaces, which are the network interfaces that connect the notebook instances to the VPC. The other options are not correct because: Option B: Creating an IAM policy that allows the sa gemaker:CreatePresignedNotebookInstanceUrl and sagemaker:DescribeNotebookInstance actions from only the VPC endpoints does not prevent individuals outside the VPC from accessing the note book instances. These actions are used to generate and retrieve the authorized URL for the no tebook instances, but they do not control who can use the URL to access the notebook instances. T he URL can still be shared or leaked to unauthorized users, who can then access the noteboo k instances over the internet. Option C: Adding a NAT gateway to the VPC and conve rting the subnets where the notebook instances are hosted to private subnets does not so lve the issue either. A NAT gateway is used to enable outbound internet access from a private subn et, but it does not affect inbound internet access. The notebook instances can still be accesse d over the internet if their security group allows inbound traffic from any source IP address. Moreove r, stopping and starting the notebook instances to reassign only private IP addresses is not necess ary, because the notebook instances already have private IP addresses assigned by the VPC interface endpoints. Option D: Changing the network ACL of the subnet th e notebook is hosted in to restrict access to anyone outside the VPC is not a good practice, beca use network ACLs are stateless and apply to the entire subnet. This means that the data science tea m would have to specify both the inbound and outbound rules for each IP address range that they want to allow or deny. This can be cumbersome and error-prone, especially if the VPC has multiple subnets and resources. It is better to use securit y groups, which are stateful and apply to individual resources, to control the access to the notebook instances.", + "references": "Connect to SageMaker Within your VPC - Amazon SageM aker Security Groups for Your VPC - Amazon Virtual Priva te Cloud VPC Interface Endpoints - Amazon Virtual Private Cl oud" + }, + { + "question": "A company will use Amazon SageMaker to train and ho st a machine learning (ML) model for a marketing campaign. The majority of data is sensiti ve customer dat", + "options": [ + "A. The data must be encrypted at rest. The company w ants AWS to maintain the root of trust for the", + "B. Use encryption keys that are stored in AWS Cloud HSM to encrypt the ML data volumes, and to", + "C. Use SageMaker built-in transient keys to encrypt the ML data volumes. Enable default encryption", + "D. Use customer managed keys in AWS Key Management S ervice (AWS KMS) to encrypt the ML data" + ], + "correct": "", + "explanation": "Amazon SageMaker supports encryption at rest for th e ML storage volumes, the model artifacts, and the data in Amazon S3 using AWS Key Management Serv ice (AWS KMS). AWS KMS is a service that allows customers to create and manage encryption ke ys that can be used to encrypt data. AWS KMS also provides an audit trail of key usage by loggin g key events to AWS CloudTrail. Customers can use either AWS managed keys or customer managed keys to encrypt their data. AWS managed keys are created and managed by AWS on behalf of the custome r, while customer managed keys are created and managed by the customer. Customer managed keys offer more control and flexibility over the key policies, permissions, and rotation. Therefore, to meet the requirements of the company, the best option is to use customer managed keys in AWS KMS to encrypt the ML data volumes, and to encrypt the model artifacts and data in Amazon S3. The other options are not correct because: Option A: AWS Cloud HSM is a service that provides hardware security modules (HSMs) to store and use encryption keys. AWS Cloud HSM is not integrate d with Amazon SageMaker, and cannot be used to encrypt the ML data volumes, the model artifacts , or the data in Amazon S3. AWS Cloud HSM is more suitable for customers who need to meet strict compliance requirements or who need direct control over the HSMs. Option B: SageMaker built-in transient keys are tem porary keys that are used to encrypt the ML data volumes and are discarded immediately after encrypt ion. These keys do not provide persistent encryption or logging of key usage. Enabling defaul t encryption for new Amazon Elastic Block Store (Amazon EBS) volumes does not affect the ML data vo lumes, which are encrypted separately by SageMaker. Moreover, this option does not address t he encryption of the model artifacts and data in Amazon S3. Option D: AWS Security Token Service (AWS STS) is a service that provides temporary credentials to access AWS resources. AWS STS does not provide encr yption keys or encryption services. AWS STS cannot be used to encrypt the ML storage volumes, t he model artifacts, or the data in Amazon S3.", + "references": "Protect Data at Rest Using Encryption - Amazon Sage Maker What is AWS Key Management Service? - AWS Key Manag ement Service What is AWS CloudHSM? - AWS CloudHSM What is AWS Security Token Service? - AWS Security Token Service" + }, + { + "question": "A machine learning specialist stores IoT soil senso r data in Amazon DynamoDB table and stores weather event data as JSON files in Amazon S3. The dataset in DynamoDB is 10 GB in size and the dataset in Amazon S3 is 5 GB in size. The specialis t wants to train a model on this data to help predi ct soil moisture levels as a function of weather event s using Amazon SageMaker. Which solution will accomplish the necessary transf ormation to train the Amazon SageMaker model with the LEAST amount of administrative overhead?", + "options": [ + "A. Launch an Amazon EMR cluster. Create an Apache Hi ve external table for the DynamoDB table and", + "B. Crawl the data using AWS Glue crawlers. Write an AWS Glue ETL job that merges the two tables", + "C. Enable Amazon DynamoDB Streams on the sensor tabl e. Write an AWS Lambda function that", + "D. Crawl the data using AWS Glue crawlers. Write an AWS Glue ETL job that merges the two tables" + ], + "correct": "D. Crawl the data using AWS Glue crawlers. Write an AWS Glue ETL job that merges the two tables", + "explanation": "Explanation/Reference: The solution that will accomplish the necessary tra nsformation to train the Amazon SageMaker model with the least amount of administrative overh ead is to crawl the data using AWS Glue crawlers, write an AWS Glue ETL job that merges the two tables and writes the output in CSV format to Amazon S3. This solution leverages the serverles s capabilities of AWS Glue to automatically discover the schema of the data sources, and to per form the data integration and transformation without requiring any cluster management or configu ration. The output in CSV format is compatible with Amazon SageMaker and can be easily loaded into a training job. Reference: AWS Glue, Amazon SageMaker", + "references": "" + }, + { + "question": "A company sells thousands of products on a public w ebsite and wants to automatically identify products with potential durability problems. The co mpany has 1.000 reviews with date, star rating, review text, review summary, and customer email fie lds, but many reviews are incomplete and have empty fields. Each review has already been labeled with the correct durability result. A machine learning specialist must train a model to identify reviews expressing concerns over product durability. The first model needs to be tra ined and ready to review in 2 days. What is the MOST direct approach to solve this prob lem within 2 days?", + "options": [ + "A. Train a custom classifier by using Amazon Compreh end.", + "B. Build a recurrent neural network (RNN) in Amazon SageMaker by using Gluon and Apache MXNet.", + "C. Train a built-in BlazingText model using Word2Vec mode in Amazon SageMaker.", + "D. Use a built-in seq2seq model in Amazon SageMaker." + ], + "correct": "A. Train a custom classifier by using Amazon Compreh end.", + "explanation": "The most direct approach to solve this problem with in 2 days is to train a custom classifier by using Amazon Comprehend. Amazon Comprehend is a natural l anguage processing (NLP) service that can analyze text and extract insights such as sentiment , entities, topics, and syntax. Amazon Comprehend also provides a custom classification feature that allows users to create and train a custom text classifier using their own labeled data. The custom classifier can then be used to categorize any text document into one or more custom classes. For this use case, the custom classifier can be trained to identify reviews that express concerns over product durability as a class, and use the star rating, review text, and review summary fields as input fea tures. The custom classifier can be created and trained using the Amazon Comprehend console or API, and does not require any coding or machine learning expertise. The training process is fully m anaged and scalable, and can handle large and complex datasets. The custom classifier can be trai ned and ready to review in 2 days or less, depending on the size and quality of the dataset. The other options are not the most direct approache s because: Option B: Building a recurrent neural network (RNN) in Amazon SageMaker by using Gluon and Apache MXNet is a more complex and time-consuming a pproach that requires coding and machine learning skills. RNNs are a type of deep learning m odels that can process sequential data, such as text, and learn long-term dependencies between toke ns. Gluon is a high-level API for MXNet that simplifies the development of deep learning models. Amazon SageMaker is a fully managed service that provides tools and frameworks for building, tr aining, and deploying machine learning models. However, to use this approach, the machine learning specialist would have to write custom code to preprocess the data, define the RNN architecture, t rain the model, and evaluate the results. This would likely take more than 2 days and involve more administrative overhead. Option C: Training a built-in BlazingText model usi ng Word2Vec mode in Amazon SageMaker is not a suitable approach for text classification. BlazingT ext is a built-in algorithm in Amazon SageMaker tha t provides highly optimized implementations of the Wo rd2Vec and text classification algorithms. The Word2Vec algorithm is useful for generating word em beddings, which are dense vector representations of words that capture their semanti c and syntactic similarities. However, word embeddings alone are not sufficient for text classi fication, as they do not account for the context an d structure of the text documents. To use this approa ch, the machine learning specialist would have to combine the word embeddings with another classifier model, such as a logistic regression or a neural network, which would add more complexity and time t o the solution. Option D: Using a built-in seq2seq model in Amazon SageMaker is not a relevant approach for text classification. Seq2seq is a built-in algorithm in Amazon SageMaker that provides a sequence-tosequenc e framework for neural machine translation based on M XNet. Seq2seq is a supervised learning algorithm that can generate an output sequ ence of tokens given an input sequence of tokens, such as translating a sentence from one lan guage to another. However, seq2seq is not designed for text classification, which requires as signing a label or a category to a text document, n ot generating another text sequence. To use this appro ach, the machine learning specialist would have to modify the seq2seq algorithm to fit the text cla ssification task, which would be challenging and inefficient.", + "references": "Custom Classification - Amazon Comprehend Build a Text Classification Model with Amazon Compr ehend - AWS Machine Learning Blog Recurrent Neural Networks - Gluon API BlazingText Algorithm - Amazon SageMaker Sequence-to-Sequence Algorithm - Amazon SageMaker" + }, + { + "question": "A company that runs an online library is implementi ng a chatbot using Amazon Lex to provide book recommendations based on category. This intent is f ulfilled by an AWS Lambda function that queries an Amazon DynamoDB table for a list of book titles, given a particular category. For testing, there ar e only three categories implemented as the custom slo t types: \"comedy,\" \"adventure, and \"documentary. A machine learning (ML) specialist notices that som etimes the request cannot be fulfilled because Amazon Lex cannot understand the category spoken by users with utterances such as \"funny,\" \"fun,\" and \"humor.\" The ML specialist needs to fix the pro blem without changing the Lambda code or data in DynamoDB. How should the ML specialist fix the problem?", + "options": [ + "A. Add the unrecognized words in the enumeration val ues list as new values in the slot type.", + "B. Create a new custom slot type, add the unrecogniz ed words to this slot type as enumeration", + "C. Use the AMAZON.SearchQuery built-in slot types fo r custom searches in the database.", + "D. Add the unrecognized words as synonyms in the cus tom slot type." + ], + "correct": "D. Add the unrecognized words as synonyms in the cus tom slot type.", + "explanation": "The best way to fix the problem without changing th e Lambda code or data in DynamoDB is to add the unrecognized words as synonyms in the custom sl ot type. This way, Amazon Lex can resolve the synonyms to the corresponding slot values and pass them to the Lambda function. For example, if the slot type has a value oecomedy with synonyms oe funny , oefun , and oehumor , then any of these words entered by the user will be resolved to oecom edy and the Lambda function can query the DynamoDB table for the book titles in that category . Adding synonyms to the custom slot type can be done easily using the Amazon Lex console or API, an d does not require any code changes. The other options are not correct because: Option A: Adding the unrecognized words in the enum eration values list as new values in the slot type would not fix the problem, because the Lambda function and the DynamoDB table are not aware of these new values. The Lambda function woul d not be able to query the DynamoDB table for the book titles in the new categories, and the requ est would still fail. Moreover, adding new values t o the slot type would increase the complexity and mai ntenance of the chatbot, as the Lambda function and the DynamoDB table would have to be updated acc ordingly. Option B: Creating a new custom slot type, adding t he unrecognized words to this slot type as enumeration values, and using this slot type for th e slot would also not fix the problem, for the same reasons as option A. The Lambda function and the DynamoDB table would not be able to handle the new slot type and its values, and the request would still fail. Furthermore, creating a new slot type would require more effort and time than adding syno nyms to the existing slot type. Option C: Using the AMAZON.SearchQuery built-in slo t types for custom searches in the database is not a suitable approach for this use case. The AMAZ ON.SearchQuery slot type is used to capture freefor m user input that corresponds to a search query. Howe ver, this slot type does not perform any validation or resolution of the user input, and pas ses the raw input to the Lambda function. This means that the Lambda function would have to handle the logic of parsing and matching the user input to the DynamoDB table, which would require ch anging the Lambda code and adding more complexity to the solution.", + "references": "Custom slot type - Amazon Lex Using Synonyms - Amazon Lex Built-in Slot Types - Amazon Lex" + }, + { + "question": "A manufacturing company uses machine learning (ML) models to detect quality issues. The models use images that are taken of the company's product at the end of each production step. The company has thousands of machines at the production site th at generate one image per second on average. The company ran a successful pilot with a single ma nufacturing machine. For the pilot, ML specialists used an industrial PC that ran AWS IoT Greengrass w ith a long-running AWS Lambda function that uploaded the images to Amazon S3. The uploaded imag es invoked a Lambda function that was written in Python to perform inference by using an Amazon SageMaker endpoint that ran a custom model. The inference results were forwarded back to a web service that was hosted at the production site to prevent faulty products from bei ng shipped. The company scaled the solution out to all manufact uring machines by installing similarly configured industrial PCs on each production machine. However, latency for predictions increased beyond acceptable limits. Analysis shows that the internet connection is at its capacity limit. How can the company resolve this issue MOST cost-ef fectively?", + "options": [ + "A. Set up a 10 Gbps AWS Direct Connect connection be tween the production site and the nearest", + "B. Extend the long-running Lambda function that runs on AWS IoT Greengrass to compress the", + "C. Use auto scaling for SageMaker. Set up an AWS Dir ect Connect connection between the", + "D. Deploy the Lambda function and the ML models onto the AWS IoT Greengrass core that is running" + ], + "correct": "D. Deploy the Lambda function and the ML models onto the AWS IoT Greengrass core that is running", + "explanation": "The best option is to deploy the Lambda function an d the ML models onto the AWS IoT Greengrass core that is running on the industrial PCs that are installed on each machine. This way, the inference can be performed locally on the edge devices, witho ut the need to upload the images to Amazon S3 and invoke the SageMaker endpoint. This will reduce the latency and the network bandwidth consumption. The long-running Lambda function can b e extended to invoke the Lambda function with the captured images and run the inference on t he edge component that forwards the results directly to the web service. This will also simplif y the architecture and eliminate the dependency on the internet connection. Option A is not cost-effective, as it requires sett ing up a 10 Gbps AWS Direct Connect connection and increasing the size and number of instances for the SageMaker endpoint. This will increase the operational costs and complexity. Option B is not optimal, as it still requires uploa ding the images to Amazon S3 and invoking the SageMaker endpoint. Compressing and decompressing t he images will add additional processing overhead and latency. Option C is not sufficient, as it still requires up loading the images to Amazon S3 and invoking the SageMaker endpoint. Auto scaling for SageMaker will help to handle the increased workload, but it will not reduce the latency or the network bandwidt h consumption. Setting up an AWS Direct Connect connection will improve the network perform ance, but it will also increase the operational costs and complexity. Reference: AWS IoT Greengrass Deploying Machine Learning Models to Edge Devices AWS Certified Machine Learning - Specialty Exam Gui de", + "references": "" + }, + { + "question": "A data scientist is using an Amazon SageMaker noteb ook instance and needs to securely access data stored in a specific Amazon S3 bucket. How should the data scientist accomplish this?", + "options": [ + "A. Add an S3 bucket policy allowing GetObject, PutOb ject, and ListBucket permissions to the Amazon", + "B. Encrypt the objects in the S3 bucket with a custo m AWS Key Management Service (AWS KMS) key", + "C. Attach the policy to the IAM role associated with the notebook that allows GetObject, PutObject,", + "D. Use a script in a lifecycle configuration to conf igure the AWS CLI on the instance with an access ke y" + ], + "correct": "C. Attach the policy to the IAM role associated with the notebook that allows GetObject, PutObject,", + "explanation": "The best way to securely access data stored in a sp ecific Amazon S3 bucket from an Amazon SageMaker notebook instance is to attach a policy t o the IAM role associated with the notebook that allows GetObject, PutObject, and ListBucket operati ons to the specific S3 bucket. This way, the notebook can use the AWS SDK or CLI to access the S 3 bucket without exposing any credentials or requiring any additional configuration. This is als o the recommended approach by AWS for granting access to S3 from SageMaker. Reference: Amazon SageMaker Roles Accessing Amazon S3 from a SageMaker Notebook Insta nce", + "references": "" + }, + { + "question": "A company is launching a new product and needs to b uild a mechanism to monitor comments about the company and its new product on social medi", + "options": [ + "A. The company needs to be able to evaluate the sent iment expressed in social media posts, and", + "B. Train a model in Amazon SageMaker by using the Bl azingText algorithm to detect sentiment in the", + "D. Trigger an AWS Lambda function when social media posts are added to the S3 bucket. Call Amazon" + ], + "correct": "D. Trigger an AWS Lambda function when social media posts are added to the S3 bucket. Call Amazon", + "explanation": "The solution that uses Amazon Comprehend and Amazon CloudWatch is the most suitable for the given scenario. Amazon Comprehend is a natural lang uage processing (NLP) service that can analyze text and extract insights such as sentiment, entiti es, topics, and syntax. Amazon CloudWatch is a monitoring and observability service that can colle ct and track metrics, create dashboards, and set alarms based on various thresholds. By using these services, the data science team can quickly and easily implement a solution to monitor the sentimen t of social media posts without requiring much infrastructure or data science resources. The solut ion also meets the requirements of storing the sentiment in both S3 and CloudWatch, and using Clou dWatch alarms to notify analysts of trends.", + "references": "Amazon Comprehend Amazon CloudWatch" + }, + { + "question": "A bank wants to launch a low-rate credit promotion. The bank is located in a town that recently experienced economic hardship. Only some of the ban k's customers were affected by the crisis, so the bank's credit team must identify which customer s to target with the promotion. However, the credit team wants to make sure that loyal customers ' full credit history is considered when the decision is made. The bank's data science team developed a model that classifies account transactions and understands credit eligibility. The data science te am used the XGBoost algorithm to train the model. The team used 7 years of bank transaction historica l data for training and hyperparameter tuning over the course of several days. The accuracy of the model is sufficient, but the cr edit team is struggling to explain accurately why t he model denies credit to some customers. The credit t eam has almost no skill in data science. What should the data science team do to address thi s issue in the MOST operationally efficient manner?", + "options": [ + "A. Use Amazon SageMaker Studio to rebuild the model. Create a notebook that uses the XGBoost", + "B. Use Amazon SageMaker Studio to rebuild the model. Create a notebook that uses the XGBoost", + "C. Create an Amazon SageMaker notebook instance. Use the notebook instance and the XGBoost library to locally retrain the model. Use the plot_ importance() method in the Python XGBoost", + "D. Use Amazon SageMaker Studio to rebuild the model. Create a notebook that uses the XGBoost" + ], + "correct": "A. Use Amazon SageMaker Studio to rebuild the model. Create a notebook that uses the XGBoost", + "explanation": "The best option is to use Amazon SageMaker Studio t o rebuild the model and deploy it at an endpoint. Then, use Amazon SageMaker Model Monitor to store inferences and use the inferences to create Shapley values that help explain model behav ior. Shapley values are a way of attributing the contribution of each feature to the model output. T hey can help the credit team understand why the model makes certain decisions and how the features affect the model outcomes. A chart that shows features and SHapley Additive exPlanations (SHAP) v alues can be created using the SHAP library in Python. This option is the most operationally effic ient because it leverages the existing XGBoost training container and the built-in capabilities of Amazon SageMaker Model Monitor and SHAP library. Reference: Amazon SageMaker Studio Amazon SageMaker Model Monitor SHAP library", + "references": "" + }, + { + "question": "A data science team is planning to build a natural language processing (NLP) application. The applications text preprocessing stage will include part-of-speech tagging and key phase extraction. The preprocessed text will be input to a custom cla ssification algorithm that the data science team has already written and trained using Apache MXNet. Which solution can the team build MOST quickly to m eet these requirements?", + "options": [ + "A. Use Amazon Comprehend for the part-of-speech tagg ing, key phase extraction, and classification", + "B. Use an NLP library in Amazon SageMaker for the pa rt-of-speech tagging. Use Amazon", + "C. Use Amazon Comprehend for the part-of-speech tagg ing and key phase extraction tasks. Use", + "D. Use Amazon Comprehend for the part-of-speech tagg ing and key phase extraction tasks. Use AWS" + ], + "correct": "D. Use Amazon Comprehend for the part-of-speech tagg ing and key phase extraction tasks. Use AWS", + "explanation": "Amazon Comprehend is a natural language processing (NLP) service that can perform part-of-speech tagging and key phrase extraction tasks. AWS Deep L earning Containers are Docker images that are pre-installed with popular deep learning frameworks such as Apache MXNet. Amazon SageMaker is a fully managed service that can help build, train, a nd deploy machine learning models. Using Amazon Comprehend for the text preprocessing tasks and AWS Deep Learning Containers with Amazon SageMaker to build the custom classifier is the sol ution that can be built most quickly to meet the requirements.", + "references": "Amazon Comprehend AWS Deep Learning Containers Amazon SageMaker" + }, + { + "question": "A machine learning (ML) specialist must develop a c lassification model for a financial services company. A domain expert provides the dataset, whic h is tabular with 10,000 rows and 1,020 features. During exploratory data analysis, the spe cialist finds no missing values and a small percentage of duplicate rows. There are correlation scores of > 0.9 for 200 feature pairs. The mean value of each feature is similar to its 50th percen tile. Which feature engineering strategy should the ML sp ecialist use with Amazon SageMaker?", + "options": [ + "A. Apply dimensionality reduction by using the princ ipal component analysis (PCA) algorithm.", + "B. Drop the features with low correlation scores by using a Jupyter notebook.", + "C. Apply anomaly detection by using the Random Cut F orest (RCF) algorithm.", + "D. Concatenate the features with high correlation sc ores by using a Jupyter notebook." + ], + "correct": "A. Apply dimensionality reduction by using the princ ipal component analysis (PCA) algorithm.", + "explanation": "The best feature engineering strategy for this scen ario is to apply dimensionality reduction by using the principal component analysis (PCA) algorithm. P CA is a technique that transforms a large set of correlated features into a smaller set of uncorrela ted features called principal components. This can help reduce the complexity and noise in the data, i mprove the performance and interpretability of the model, and avoid overfitting. Amazon SageMaker provides a built-in PCA algorithm that can be used to perform dimensionality reduction on tabular data. The ML specialist can use Amazon SageMaker to train and deploy the PCA model, and th en use the output of the PCA model as the input for the classification model.", + "references": "Dimensionality Reduction with Amazon SageMaker Amazon SageMaker PCA Algorithm" + }, + { + "question": "A machine learning specialist needs to analyze comm ents on a news website with users across the globe. The specialist must find the most discussed topics in the comments that are in either English or Spanish. What steps could be used to accomplish this task? ( Choose two.)", + "options": [ + "A. Use an Amazon SageMaker BlazingText algorithm to find the topics independently from language.", + "B. Use an Amazon SageMaker seq2seq algorithm to tran slate from Spanish to English, if necessary.", + "C. Use Amazon Translate to translate from Spanish to English, if necessary. Use Amazon Comprehend", + "D. Use Amazon Translate to translate from Spanish to English, if necessary. Use Amazon Lex to extract" + ], + "correct": "", + "explanation": "Explanation/Reference: To find the most discussed topics in the comments t hat are in either English or Spanish, the machine learning specialist needs to perform two steps: fir st, translate the comments from Spanish to English if necessary, and second, apply a topic modeling al gorithm to the comments. The following options are valid ways to accomplish these steps using AWS services: Option C: Use Amazon Translate to translate from Sp anish to English, if necessary. Use Amazon Comprehend topic modeling to find the topics. Amazo n Translate is a neural machine translation service that delivers fast, high-quality, and affor dable language translation. Amazon Comprehend is a natural language processing (NLP) service that uses machine learning to find insights and relationships in text. Amazon Comprehend topic mode ling is a feature that automatically organizes a collection of text documents into topics that conta in commonly used words and phrases. Option E: Use Amazon Translate to translate from Sp anish to English, if necessary. Use Amazon SageMaker Neural Topic Model (NTM) to find the topi cs. Amazon SageMaker is a fully managed service that provides every developer and data scie ntist with the ability to build, train, and deploy machine learning (ML) models quickly. Amazon SageMa ker Neural Topic Model (NTM) is an unsupervised learning algorithm that is used to org anize a corpus of documents into topics that contain word groupings based on their statistical d istribution. The other options are not valid because: Option A: Amazon SageMaker BlazingText algorithm is not a topic modeling algorithm, but a text classification and word embedding algorithm. It can not find the topics independently from language, as different languages have different word distribu tions and semantics. Option B: Amazon SageMaker seq2seq algorithm is not a translation algorithm, but a sequence-tosequence learning algorithm that can be used for tasks such as summarization, chatbot, and question answering. Amazon SageMaker Latent Dirichlet Alloca tion (LDA) algorithm is a topic modeling algorithm, but it requires the input documents to b e in the same language and preprocessed into a bag-of-words format. Option D: Amazon Lex is not a topic modeling algori thm, but a service for building conversational interfaces into any application using voice and tex t. It cannot extract topics from the content, but only intents and slots based on a predefined bot co nfiguration. Reference: Amazon Translate Amazon Comprehend Amazon SageMaker Amazon SageMaker Neural Topic Model (NTM) Algorithm Amazon SageMaker BlazingText Amazon SageMaker Seq2Seq Amazon SageMaker Latent Dirichlet Allocation (LDA) Algorithm Amazon Lex", + "references": "" + }, + { + "question": "A machine learning (ML) specialist is administering a production Amazon SageMaker endpoint with model monitoring configured. Amazon SageMaker Model Monitor detects violations on the SageMaker endpoint, so the ML specialist retrains t he model with the latest dataset. This dataset is statistically representative of the current product ion traffic. The ML specialist notices that even af ter deploying the new SageMaker model and running the f irst monitoring job, the SageMaker endpoint still has violations. What should the ML specialist do to resolve the vio lations?", + "options": [ + "A. Manually trigger the monitoring job to re-evaluat e the SageMaker endpoint traffic sample.", + "B. Run the Model Monitor baseline job again on the n ew training set. Configure Model Monitor to", + "C. Delete the endpoint and recreate it with the orig inal configuration.", + "D. Retrain the model again by using a combination of the original training set and the new training" + ], + "correct": "B. Run the Model Monitor baseline job again on the n ew training set. Configure Model Monitor to", + "explanation": "Explanation/Reference: The ML specialist should run the Model Monitor base line job again on the new training set and configure Model Monitor to use the new baseline. Th is is because the baseline job computes the statistics and constraints for the data quality and model quality metrics, which are used to detect violations. If the training set changes, the baseli ne job should be updated accordingly to reflect the new distribution of the data and the model performa nce. Otherwise, the old baseline may not be representative of the current production traffic an d may cause false alarms or miss violations. Reference: Monitor data and model quality - Amazon SageMaker Detecting and analyzing incorrect model predictions with Amazon SageMaker Model Monitor and Debugger | AWS Machine Learning Blog", + "references": "" + }, + { + "question": "A company supplies wholesale clothing to thousands of retail stores. A data scientist must create a model that predicts the daily sales volume for each item for each store. The data scientist discovers that more than half of the stores have been in busi ness for less than 6 months. Sales data is highly consistent from week to week. Daily data from the d atabase has been aggregated weekly, and weeks with no sales are omitted from the current dataset. Five years (100 MB) of sales data is available in Amazon S3. Which factors will adversely impact the performance of the forecast model to be developed, and which actions should the data scientist take to mit igate them? (Choose two.)", + "options": [ + "A. Detecting seasonality for the majority of stores will be an issue. Request categorical data to relat e", + "B. The sales data does not have enough variance. Req uest external sales data from other industries", + "C. Sales data is aggregated by week. Request daily s ales data from the source database to enable", + "D. The sales data is missing zero entries for item s ales. Request that item sales data from the source" + ], + "correct": "", + "explanation": "Explanation/Reference: The factors that will adversely impact the performa nce of the forecast model are: Sales data is aggregated by week. This will reduce the granularity and resolution of the data, and make it harder to capture the daily patterns and va riations in sales volume. The data scientist should request daily sales data from the source database t o enable building a daily model, which will be more accurate and useful for the prediction task. Sales data is missing zero entries for item sales. This will introduce bias and incompleteness in the data, and make it difficult to account for the item s that have no demand or are out of stock. The data scientist should request that item sales data from the source database include zero entries to enable building the model, which will be more robust and r ealistic. The other options are not valid because: Detecting seasonality for the majority of stores wi ll not be an issue, as sales data is highly consist ent from week to week. Requesting categorical data to r elate new stores with similar stores that have more historical data may not improve the model perf ormance significantly, and may introduce unnecessary complexity and noise. The sales data does not need to have more variance, as it reflects the actual demand and behavior of the customers. Requesting external sales data from other industries will not improve the models ability to generalize, but may introduce irrelevant and misleading information. Only 100 MB of sales data is not a problem, as it i s sufficient to train a forecast model with Amazon S3 and Amazon Forecast. Requesting 10 years of sale s data will not provide much benefit, as it may contain outdated and obsolete information that does not reflect the current market trends and customer preferences. Reference: Amazon Forecast Forecasting: Principles and Practice", + "references": "" + }, + { + "question": "An ecommerce company is automating the categorizati on of its products based on images. A data scientist has trained a computer vision model using the Amazon SageMaker image classification algorithm. The images for each product are classifi ed according to specific product lines. The accuracy of the model is too low when categorizing new products. All of the product images have the same dimensions and are stored within an Amazon S3 bucket. The company wants to improve the model so it can be used for new products as soon as possible. Which steps would improve the accuracy of the solut ion? (Choose three.)", + "options": [ + "A. Use the SageMaker semantic segmentation algorithm to train a new model to achieve improved", + "B. Use the Amazon Rekognition DetectLabels API to cl assify the products in the dataset.", + "C. Augment the images in the dataset. Use open-sourc e libraries to crop, resize, flip, rotate, and", + "D. Use a SageMaker notebook to implement the normali zation of pixels and scaling of the images." + ], + "correct": "", + "explanation": "Option C is correct because augmenting the images i n the dataset can help the model learn more features and generalize better to new products. Ima ge augmentation is a common technique to increase the diversity and size of the training dat a. Option E is correct because Amazon Rekognition Cust om Labels can train a custom model to detect specific objects and scenes that are relevant to th e business use case. It can also leverage the exist ing models from Amazon Rekognition that are trained on tens of millions of images across many categories. Option F is correct because class imbalance can aff ect the performance and accuracy of the model, as it can cause the model to be biased towards the maj ority class and ignore the minority class. Applying oversampling or undersampling can help bal ance the classes and improve the models ability to learn from the data. Option A is incorrect because the semantic segmenta tion algorithm is used to assign a label to every pixel in an image, not to classify the whole image into a category. Semantic segmentation is useful for applications such as autonomous driving, medica l imaging, and satellite imagery analysis. Option B is incorrect because the DetectLabels API is a general-purpose image analysis service that can detect objects, scenes, and concepts in an imag e, but it cannot be customized to the specific product lines of the ecommerce company. The DetectL abels API is based on the pre-trained models from Amazon Rekognition, which may not cover all th e categories that the company needs. Option D is incorrect because normalizing the pixel s and scaling the images are preprocessing steps that should be done before training the model, not after. These steps can help improve the models convergence and performance, but they are not suffi cient to increase the accuracy of the model on new products.", + "references": ": Image Augmentation - Amazon SageMaker : Amazon Rekognition Custom Labels Features : [Handling Imbalanced Datasets in Machine Learning ] : [Semantic Segmentation - Amazon SageMaker] : [DetectLabels - Amazon Rekognition] : [Image Classification - MXNet - Amazon SageMaker] : [https://towardsdatascience.com/handling-imbalanc ed-datasets-in-machine-learning- 7a0e84220f28] : [https://docs.aws.amazon.com/sagemaker/latest/dg/ semantic-segmentation.html] : [https://docs.aws.amazon.com/rekognition/latest/d g/API_DetectLabels.html] : [https://docs.aws.amazon.com/sagemaker/latest/dg/ image-classification.html] : [https://towardsdatascience.com/handling-imbalanc ed-datasets-in-machine-learning- 7a0e84220f28] : [https://docs.aws.amazon.com/sagemaker/latest/dg/ semantic-segmentation.html] : [https://docs.aws.amazon.com/rekognition/latest/d g/API_DetectLabels.html] : [https://docs.aws.amazon.com/sagemaker/latest/dg/ image-classification.html] : [https://towardsdatascience.com/handling-imbalanc ed-datasets-in-machine-learning- 7a0e84220f28] : [https://docs.aws.amazon.com/sagemaker/latest/dg/ semantic-segmentation.html] : [https://docs.aws.amazon.com/rekognition/latest/d g/API_DetectLabels.html] : [https://docs.aws.amazon.com/sagemaker/latest/dg/ image-classification.html]" + }, + { + "question": "A data scientist is training a text classification model by using the Amazon SageMaker built-in BlazingText algorithm. There are 5 classes in the d ataset, with 300 samples for category A, 292 samples for category B, 240 samples for category C, 258 samples for category D, and 310 samples for category E. The data scientist shuffles the data and splits off 10% for testing. After training the model, the dat a scientist generates confusion matrices for the trai ning and test sets. What could the data scientist conclude form these r esults?", + "options": [ + "A. Classes C and D are too similar.", + "B. The dataset is too small for holdout cross-valida tion.", + "C. The data distribution is skewed.", + "D. The model is overfitting for classes B and E." + ], + "correct": "D. The model is overfitting for classes B and E.", + "explanation": "A confusion matrix is a matrix that summarizes the performance of a machine learning model on a set of test data. It displays the number of true po sitives (TP), true negatives (TN), false positives (FP), and false negatives (FN) produced by the model on t he test data1. For multi-class classification, the matrix shape will be equal to the number of classes i.e for n classes it will be nXn1. The diagonal values represent the number of correct predictions for each class, and the off-diagonal values represent the number of incorrect predictions for e ach class1. The BlazingText algorithm is a proprietary machine learning algorithm for forecasting time series using causal convolutional neural networks (CNNs). BlazingText works best with large datasets containing hundreds of time series. It accepts item metadata, and is the only Forecast algorithm that accepts related time series data without future val ues2. From the confusion matrices for the training and te st sets, we can observe the following: The model has a high accuracy on the training set, as most of the diagonal values are high and the off-diagonal values are low. This means that the mo del is able to learn the patterns and features of the training data well. However, the model has a lower accuracy on the test set, as some of the diagonal values are lower and some of the off-diagonal values are higher. Thi s means that the model is not able to generalize well to the unseen data and makes more errors. The model has a particularly high error rate for cl asses B and E on the test set, as the values of M_2 2 and M_55 are much lower than the values of M_12, M_ 21, M_15, M_25, M_51, and M_52. This means that the model is confusing classes B and E w ith other classes more often than it should. The model has a relatively low error rate for class es A, C, and D on the test set, as the values of M_11, M_33, and M_44 are high and the values of M_1 3, M_14, M_23, M_24, M_31, M_32, M_34, M_41, M_42, and M_43 are low. This means that the m odel is able to distinguish classes A, C, and D from other classes well. These results indicate that the model is overfittin g for classes B and E, meaning that it is memorizin g the specific features of these classes in the train ing data, but failing to capture the general featur es that are applicable to the test data. Overfitting i s a common problem in machine learning, where the model performs well on the training data, but poorl y on the test data3. Some possible causes of overfitting are: The model is too complex or has too many parameters for the given data. This makes the model flexible enough to fit the noise and outliers in th e training data, but reduces its ability to general ize to new data. The data is too small or not representative of the population. This makes the model learn from a limited or biased sample of data, but fails to capt ure the variability and diversity of the population . The data is imbalanced or skewed. This makes the mo del learn from a disproportionate or uneven distribution of data, but fails to account for the minority or rare classes. Some possible solutions to prevent or reduce overfi tting are: Simplify the model or use regularization techniques . This reduces the complexity or the number of parameters of the model, and prevents it from fitti ng the noise and outliers in the data. Regularization techniques, such as L1 or L2 r egularization, add a penalty term to the loss function of the model, which shrinks the weights of the model and reduces overfitting3. Increase the size or diversity of the data. This pr ovides more information and examples for the model to learn from, and increases its ability to general ize to new data. Data augmentation techniques, such as rotation, flipping, cropping, or noise addi tion, can generate new data from the existing data by applying some transformations3. Balance or resample the data. This adjusts the dist ribution or the frequency of the data, and ensures that the model learns from all classes equally. Res ampling techniques, such as oversampling or undersampling, can create a balanced dataset by inc reasing or decreasing the number of samples for each class3.", + "references": "Confusion Matrix in Machine Learning - GeeksforGeek s BlazingText algorithm - Amazon SageMaker Overfitting and Underfitting in Machine Learning - GeeksforGeeks" + }, + { + "question": "A company that manufactures mobile devices wants to determine and calibrate the appropriate sales price for its devices. The company is collecting th e relevant data and is determining data features that it can use to train machine learning (ML) mode ls. There are more than 1,000 features, and the company wants to determine the primary features tha t contribute to the sales price. Which techniques should the company use for feature selection? (Choose three.)", + "options": [ + "A. Data scaling with standardization and normalizati on", + "B. Correlation plot with heat maps", + "C. Data binning", + "D. Univariate selection" + ], + "correct": "", + "explanation": "Feature selection is the process of selecting a sub set of extracted features that are relevant and contribute to minimizing the error rate of a traine d model. Some techniques for feature selection are: Correlation plot with heat maps: This technique vis ualizes the correlation between features using a color-coded matrix. Features that are highly correl ated with each other or with the target variable can be identified and removed to reduce redundancy and noise. Univariate selection: This technique evaluates each feature individually based on a statistical test, such as chi-square, ANOVA, or mutual information, a nd selects the features that have the highest scores or p-values. This technique is simple and fa st, but it does not consider the interactions between features. Feature importance with a tree-based classifier: Th is technique uses a tree-based classifier, such as random forest or gradient boosting, to rank the fea tures based on their importance in splitting the nodes. Features that have low importance scores can be dropped from the model. This technique can capture the non-linear relationships and interactio ns between features. The other options are not techniques for feature se lection, but rather for feature engineering, which is the process of creating, transforming, or extrac ting features from the original data. Feature engineering can improve the performance and interpr etability of the model, but it does not reduce the number of features. Data scaling with standardization and normalization : This technique transforms the features to have a common scale, such as zero mean and unit variance , or a range between 0 and 1. This technique can help some algorithms, such as k-means or logist ic regression, to converge faster and avoid numerical instability, but it does not change the n umber of features. Data binning: This technique groups the continuous features into discrete bins or categories based on some criteria, such as equal width, equal frequency , or clustering. This technique can reduce the noise and outliers in the data, and also create ord inal or nominal features that can be used for some algorithms, such as decision trees or naive Bayes, but it does not reduce the number of features. Data augmentation: This technique generates new dat a from the existing data by applying some transformations, such as rotation, flipping, croppi ng, or noise addition. This technique can increase the size and diversity of the data, and help preven t overfitting, but it does not reduce the number of features.", + "references": "Feature engineering - Machine Learning Lens Amazon SageMaker Autopilot now provides feature sel ection and the ability to change data types while creating an AutoML experiment Feature Selection in Machine Learning | Baeldung on Computer Science Feature Selection in Machine Learning: An easy Intr oduction" + }, + { + "question": "A power company wants to forecast future energy con sumption for its customers in residential properties and commercial business properties. Hist orical power consumption data for the last 10 years is available. A team of data scientists who p erformed the initial data analysis and feature selection will include the historical power consump tion data and data such as weather, number of individuals on the property, and public holidays. The data scientists are using Amazon Forecast to ge nerate the forecasts. Which algorithm in Forecast should the data scienti sts use to meet these requirements?", + "options": [ + "A. Autoregressive Integrated Moving Average (AIRMA)", + "B. Exponential Smoothing (ETS)", + "C. Convolutional Neural Network - Quantile Regressio n (CNN-QR)", + "D. Prophet" + ], + "correct": "C. Convolutional Neural Network - Quantile Regressio n (CNN-QR)", + "explanation": "CNN-QR is a proprietary machine learning algorithm for forecasting time series using causal convolutional neural networks (CNNs). CNN-QR works best with large datasets containing hundreds of time series. It accepts item metadata, and is th e only Forecast algorithm that accepts related time series data without future values. In this case, th e power company has historical power consumption data for the last 10 years, which is a large datase t with multiple time series. The data also includes related data such as weather, number of individuals on the property, and public holidays, which can be used as item metadata or related time series dat a. Therefore, CNN-QR is the most suitable algorithm for this scenario. Reference: Amazon Fore cast Algorithms, Amazon Forecast CNN-QR", + "references": "" + }, + { + "question": "A company wants to use automatic speech recognition (ASR) to transcribe messages that are less than 60 seconds long from a voicemail-style applica tion. The company requires the correct identification of 200 unique product names, some of which have unique spellings or pronunciations. The company has 4,000 words of Amazon SageMaker Gro und Truth voicemail transcripts it can use to customize the chosen ASR model. The company needs t o ensure that everyone can update their customizations multiple times each hour. Which approach will maximize transcription accuracy during the development phase?", + "options": [ + "A. Use a voice-driven Amazon Lex bot to perform the ASR customization. Create customer slots", + "B. Use Amazon Transcribe to perform the ASR customiz ation. Analyze the word confidence scores in", + "C. Create a custom vocabulary file containing each p roduct name with phonetic pronunciations, and", + "D. Use the audio transcripts to create a training da taset and build an Amazon Transcribe custom" + ], + "correct": "C. Create a custom vocabulary file containing each p roduct name with phonetic pronunciations, and", + "explanation": "The best approach to maximize transcription accurac y during the development phase is to create a custom vocabulary file containing each product name with phonetic pronunciations, and use it with Amazon Transcribe to perform the ASR customization. A custom vocabulary is a list of words and phrases that are likely to appear in your audio inp ut, along with optional information about how to pronounce them. By using a custom vocabulary, you c an improve the transcription accuracy of domain-specific terms, such as product names, that may not be recognized by the general vocabulary of Amazon Transcribe. You can also analyze the tran scripts and manually update the custom vocabulary file to include updated or additional en tries for those names that are not being correctly identified. The other options are not as effective as option C for the following reasons: Option A is not suitable because Amazon Lex is a se rvice for building conversational interfaces, not for transcribing voicemail messages. Amazon Lex als o has a limit of 100 slots per bot, which is not enough to accommodate the 200 unique product names required by the company. Option B is not optimal because it relies on the wo rd confidence scores in the transcript, which may not be accurate enough to identify all the mis-tran scribed product names. Moreover, automatically creating or updating a custom vocabulary file may i ntroduce errors or inconsistencies in the pronunciation or display of the words. Option D is not feasible because it requires a larg e amount of training data to build a custom language model. The company only has 4,000 words of Amazon SageMaker Ground Truth voicemail transcripts, which is not enough to train a robust and reliable custom language model. Additionally, creating and updating a custom language model is a time-consuming and resource-intensive process, which may not be suitable for the development phase where frequent changes are expected.", + "references": "Amazon Transcribe \" Custom Vocabulary Amazon Transcribe \" Custom Language Models [Amazon Lex \" Limits]" + }, + { + "question": "A company is building a demand forecasting model ba sed on machine learning (ML). In the development stage, an ML specialist uses an Amazon SageMaker notebook to perform feature engineering during work hours that consumes low amo unts of CPU and memory resources. A data engineer uses the same notebook to perform data pre processing once a day on average that requires very high memory and completes in only 2 hours. The data preprocessing is not configured to use GPU. All the processes are running well on an ml.m5 .4xlarge notebook instance. The company receives an AWS Budgets alert that the billing for this month exceeds the allocated budget. Which solution will result in the MOST cost savings ?", + "options": [ + "A. Change the notebook instance type to a memory opt imized instance with the same vCPU number", + "B. Keep the notebook instance type and size the same . Stop the notebook when it is not in use. Run", + "C. Change the notebook instance type to a smaller ge neral-purpose instance. Stop the notebook", + "D. Change the notebook instance type to a smaller ge neral-purpose instance. Stop the notebook" + ], + "correct": "C. Change the notebook instance type to a smaller ge neral-purpose instance. Stop the notebook", + "explanation": "The best solution to reduce the cost of the noteboo k instance and the data preprocessing job is to change the notebook instance type to a smaller gene ral-purpose instance, stop the notebook when it is not in use, and run data preprocessing on an ml. r5 instance with the same memory size as the ml.m5.4xlarge instance by using Amazon SageMaker Pr ocessing. This solution will result in the most cost savings because: Changing the notebook instance type to a smaller ge neral-purpose instance will reduce the hourly cost of running the notebook, since the feature eng ineering development does not require high CPU and memory resources. For example, an ml.t3.medium instance costs $0.0464 per hour, while an ml.m5.4xlarge instance costs $0.888 per hour1. Stopping the notebook when it is not in use will al so reduce the cost, since the notebook will only incur charges when it is running. For example, if t he notebook is used for 8 hours per day, 5 days per week, then stopping it when it is not in use will s ave about 76% of the monthly cost compared to leaving it running all the time2. Running data preprocessing on an ml.r5 instance wit h the same memory size as the ml.m5.4xlarge instance by using Amazon SageMaker Processing will reduce the cost of the data preprocessing job, since the ml.r5 instance is optimized for memory-in tensive workloads and has a lower cost per GB of memory than the ml.m5 instance. For example, an ml. r5.4xlarge instance has 128 GB of memory and costs $1.008 per hour, while an ml.m5.4xlarge insta nce has 64 GB of memory and costs $0.888 per hour1. Therefore, the ml.r5.4xlarge instance can pr ocess the same amount of data in half the time and at a lower cost than the ml.m5.4xlarge instance . Moreover, using Amazon SageMaker Processing will allow the data preprocessing job to run on a s eparate, fully managed infrastructure that can be scaled up or down as needed, without affecting the notebook instance. The other options are not as effective as option C for the following reasons: Option A is not optimal because changing the notebo ok instance type to a memory optimized instance with the same vCPU number as the ml.m5.4xl arge instance has will not reduce the cost of the notebook, since the memory optimized instances have a higher cost per vCPU than the generalpurpose instances. For example, an ml.r5.4xlarge instance h as 16 vCPUs and costs $1.008 per hour, while an ml.m5.4xlarge instance has 16 vCPUs and co sts $0.888 per hour1. Moreover, running both data preprocessing and feature engineering developm ent on the same instance will not take advantage of the scalability and flexibility of Ama zon SageMaker Processing. Option B is not suitable because running data prepr ocessing on a P3 instance type with the same memory as the ml.m5.4xlarge instance by using Amazo n SageMaker Processing will not reduce the cost of the data preprocessing job, since the P3 in stance type is optimized for GPU-based workloads and has a higher cost per GB of memory than the ml. m5 or ml.r5 instance types. For example, an ml.p3.2xlarge instance has 61 GB of memory and cost s $3.06 per hour, while an ml.m5.4xlarge instance has 64 GB of memory and costs $0.888 per h our1. Moreover, the data preprocessing job does not require GPU, so using a P3 instance type w ill be wasteful and inefficient. Option D is not feasible because running data prepr ocessing on an R5 instance with the same memory size as the ml.m5.4xlarge instance by using the Reserved Instance option will not reduce the cost of the data preprocessing job, since the Reser ved Instance option requires a commitment to a consistent amount of usage for a period of 1 or 3 y ears3. However, the data preprocessing job only runs once a day on average and completes in only 2 hours, so it does not have a consistent or predictable usage pattern. Therefore, using the Res erved Instance option will not provide any cost savings and may incur additional charges for unused capacity.", + "references": "Amazon SageMaker Pricing Manage Notebook Instances - Amazon SageMaker Amazon EC2 Pricing - Reserved Instances" + }, + { + "question": "A machine learning specialist is developing a regre ssion model to predict rental rates from rental listings. A variable named Wall_Color represents th e most prominent exterior wall color of the property. The following is the sample data, excludi ng all other variables: The specialist chose a model that needs numerical i nput data. Which feature engineering approaches should the spe cialist use to allow the regression model to learn from the Wall_Color data? (Choose two.)", + "options": [ + "A. Apply integer transformation and set Red = 1, Whi te = 5, and Green = 10.", + "B. Add new columns that store one-hot representation of colors.", + "C. Replace the color name string by its length.", + "D. Create three columns to encode the color in RGB f ormat." + ], + "correct": "", + "explanation": "In this scenario, the specialist should use one-hot encoding and RGB encoding to allow the regression model to learn from the Wall_Color data. One-hot encoding is a technique used to convert categorical data into numerical data. It cr eates new columns that store one-hot representation of colors. For example, a variable n amed color has three categories: red, green, and blue. After one-hot encoding, the new variables sho uld be like this: One-hot encoding can capture the presence or absenc e of a color, but it cannot capture the intensity or hue of a color. RGB encoding is a technique used to represent colors in a digital image. It creates three columns to encode the color in RGB format. Fo r example, a variable named color has three categories: red, green, and blue. After RGB encodin g, the new variables should be like this: RGB encoding can capture the intensity and hue of a color, but it may also introduce correlation among the three columns. Therefore, using both one- hot encoding and RGB encoding can provide more information to the regression model than using either one alone.", + "references": "Feature Engineering for Categorical Data How to Perform Feature Selection with Categorical D ata" + }, + { + "question": "A data scientist is working on a public sector proj ect for an urban traffic system. While studying the traffic patterns, it is clear to the data scientist that the traffic behavior at each light is correla ted, subject to a small stochastic error term. The data scientist must model the traffic behavior to analyz e the traffic patterns and reduce congestion. How will the data scientist MOST effectively model the problem?", + "options": [ + "A. The data scientist should obtain a correlated equ ilibrium policy by formulating this problem as a", + "B. The data scientist should obtain the optimal equi librium policy by formulating this problem as a", + "C. Rather than finding an equilibrium policy, the da ta scientist should obtain accurate predictors of", + "D. Rather than finding an equilibrium policy, the da ta scientist should obtain accurate predictors of" + ], + "correct": "A. The data scientist should obtain a correlated equ ilibrium policy by formulating this problem as a", + "explanation": "The data scientist should obtain a correlated equil ibrium policy by formulating this problem as a multi-agent reinforcement learning problem. This is because: Multi-agent reinforcement learning (MARL) is a subf ield of reinforcement learning that deals with learning and coordination of multiple agents that i nteract with each other and the environment 1. MARL can be applied to problems that involve distributed decision making, such as traffic signal control, where each traffic light ca n be modeled as an agent that observes the traffic state and chooses an action (e.g., changing the sig nal phase) to optimize a reward function (e.g., minimizing the delay or congestion) 2. A correlated equilibrium is a solution concept in g ame theory that generalizes the notion of Nash equilibrium. It is a probability distribution over the joint actions of the agents that satisfies the following condition: no agent can improve its expec ted payoff by deviating from the distribution, given that it knows the distribution and the action s of the other agents 3. A correlated equilibrium can capture the correlation among the agents action s, which is useful for modeling the traffic behavior at each light that is subject to a small s tochastic error term. A correlated equilibrium policy is a policy that in duces a correlated equilibrium in a MARL setting. I t can be obtained by using various methods, such as p olicy gradient, actor-critic, or Q-learning algorithms, that can learn from the feedback of the environment and the communication among the agents 4. A correlated equilibrium policy can achie ve a better performance than a Nash equilibrium policy, which assumes that the agents act independe ntly and ignore the correlation among their actions 5. Therefore, by obtaining a correlated equilibrium po licy by formulating this problem as a MARL problem, the data scientist can most effectively mo del the traffic behavior and reduce congestion.", + "references": "Multi-Agent Reinforcement Learning Multi-Agent Reinforcement Learning for Traffic Sign al Control: A Survey Correlated Equilibrium Multi-Agent Actor-Critic for Mixed Cooperative-Comp etitive Environments Correlated Q-Learning" + }, + { + "question": "A data scientist is using the Amazon SageMaker Neur al Topic Model (NTM) algorithm to build a model that recommends tags from blog posts. The raw blog post data is stored in an Amazon S3 bucket in JSON format. During model evaluation, the data scientist discovered that the model recommends certain stopwords such as \"a,\" \"an, and \"the\" as tags to certain blog posts, along with a few rare words that are present only in certain blo g entries. After a few iterations of tag review wit h the content team, the data scientist notices that t he rare words are unusual but feasible. The data scientist also must ensure that the tag recommendat ions of the generated model do not include the stopwords. What should the data scientist do to meet these req uirements?", + "options": [ + "A. Use the Amazon Comprehend entity recognition API operations. Remove the detected words from", + "B. Run the SageMaker built-in principal component an alysis (PCA) algorithm with the blog post data", + "C. Use the SageMaker built-in Object Detection algor ithm instead of the NTM algorithm for the", + "D. Remove the stop words from the blog post data by using the Count Vectorizer function in the" + ], + "correct": "D. Remove the stop words from the blog post data by using the Count Vectorizer function in the", + "explanation": "The data scientist should remove the stop words fro m the blog post data by using the Count Vectorizer function in the scikit-learn library, an d replace the blog post data in the S3 bucket with the results of the vectorizer. This is because: The Count Vectorizer function is a tool that can co nvert a collection of text documents to a matrix of token counts 1. It also enables the pre-processing of text data prior to generating the vector representation, such as removing accents, convertin g to lowercase, and filtering out stop words 1. By using this function, the data scientist can remove the stop words such as oea, oean, and oethe from the blog post data, and obtain a numerical represen tation of the text that can be used as input for the NTM algorithm. The NTM algorithm is a neural network-based topic m odeling technique that can learn latent topics from a corpus of documents 2. It can be used to rec ommend tags from blog posts by finding the most probable topics for each document, and ranking the words associated with each topic 3. However, the NTM algorithm does not perform any text pre-pro cessing by itself, so it relies on the quality of the input data. Therefore, the data scientist shoul d replace the blog post data in the S3 bucket with the results of the vectorizer, to ensure that the N TM algorithm does not include the stop words in the tag recommendations. The other options are not suitable for the followin g reasons: Option A is not relevant because the Amazon Compreh end entity recognition API operations are used to detect and extract named entities from text , such as people, places, organizations, dates, etc4. This is not the same as removing stop words, which are common words that do not carry much meaning or information. Moreover, removing the dete cted entities from the blog post data may reduce the quality and diversity of the tag recomme ndations, as some entities may be relevant and useful as tags. Option B is not optimal because the SageMaker built -in principal component analysis (PCA) algorithm is used to reduce the dimensionality of a dataset by finding the most important features that capture the maximum amount of variance in the data 5. This is not the same as removing stop words, which are words that have low variance and h igh frequency in the data. Moreover, replacing the blog post data in the S3 bucket with the result s of the PCA algorithm may not be compatible with the input format expected by the NTM algorithm, whi ch requires a bag-of-words representation of the text 2. Option C is not suitable because the SageMaker buil t-in Object Detection algorithm is used to detect and localize objects in images 6. This is not relat ed to the task of recommending tags from blog posts , which are text documents. Moreover, using the Objec t Detection algorithm instead of the NTM algorithm would require a different type of input d ata (images instead of text), and a different type of output data (bounding boxes and labels instead of t opics and words).", + "references": "Neural Topic Model (NTM) Algorithm Introduction to the Amazon SageMaker Neural Topic M odel Amazon Comprehend - Entity Recognition sklearn.feature_extraction.text.CountVectorizer Principal Component Analysis (PCA) Algorithm Object Detection Algorithm" + }, + { + "question": "A company wants to create a data repository in the AWS Cloud for machine learning (ML) projects. The company wants to use AWS to perform complete ML lifecycles and wants to use Amazon S3 for the data storage. All of the companys data currentl y resides on premises and is 40 \u00d0\u00a2\u00d0' in size. The company wants a solution that can transfer and automatically update data between the onpremises object storage and Amazon S3. The solution must sup port encryption, scheduling, monitoring, and data integrity validation. Which solution meets these requirements?", + "options": [ + "A. Use the S3 sync command to compare the source S3 bucket and the destination S3 bucket.", + "B. Use AWS Transfer for FTPS to transfer the files f rom the on-premises storage to Amazon S3.", + "C. Use AWS DataSync to make an initial copy of the e ntire dataset. Schedule subsequent incremental", + "D. Use S3 Batch Operations to pull data periodically from the on-premises storage. Enable S3" + ], + "correct": "C. Use AWS DataSync to make an initial copy of the e ntire dataset. Schedule subsequent incremental", + "explanation": "The best solution to meet the requirements of the c ompany is to use AWS DataSync to make an initial copy of the entire dataset, and schedule su bsequent incremental transfers of changing data until the final cutover from on premises to AWS. Th is is because: AWS DataSync is an online data movement and discove ry service that simplifies data migration and helps you quickly, easily, and securely transfer yo ur file or object data to, from, and between AWS storage services 1. AWS DataSync can copy data betw een on-premises object storage and Amazon S3, and also supports encryption, scheduling, monit oring, and data integrity validation 1. AWS DataSync can make an initial copy of the entire dataset by using a DataSync agent, which is a software appliance that connects to your on-premise s storage and manages the data transfer to AWS 2. The DataSync agent can be deployed as a virt ual machine (VM) on your existing hypervisor, or as an Amazon EC2 instance in your AWS account 2. AWS DataSync can schedule subsequent incremental tr ansfers of changing data by using a task, which is a configuration that specifies the source and destination locations, the options for the transfer, and the schedule for the transfer 3. You can create a task to run once or on a recurring schedule, and you can also use filters to include o r exclude specific files or objects based on their names or prefixes 3. AWS DataSync can perform the final cutover from on premises to AWS by using a sync task, which is a type of task that synchronizes the data in the sour ce and destination locations 4. A sync task transfe rs only the data that has changed or that doesnt exist in the destination, and also deletes any files or objects from the destination that were deleted from the source since the last sync 4. Therefore, by using AWS DataSync, the company can c reate a data repository in the AWS Cloud for machine learning projects, and use Amazon S3 for th e data storage, while meeting the requirements of encryption, scheduling, monitoring, and data int egrity validation.", + "references": "Data Transfer Service - AWS DataSync Deploying a DataSync Agent Creating a Task Syncing Data with AWS DataSync" + }, + { + "question": "A company has video feeds and images of a subway tr ain station. The company wants to create a deep learning model that will alert the station man ager if any passenger crosses the yellow safety line when there is no train in the station. The ale rt will be based on the video feeds. The company wants the model to detect the yellow line, the pass engers who cross the yellow line, and the trains in the video feeds. This task requires labeling. The v ideo data must remain confidential. A data scientist creates a bounding box to label th e sample data and uses an object detection model. However, the object detection model cannot clearly demarcate the yellow line, the passengers who cross the yellow line, and the trains. Which labeling approach will help the company impro ve this model?", + "options": [ + "A. Use Amazon Rekognition Custom Labels to label the dataset and create a custom Amazon", + "B. Use an Amazon SageMaker Ground Truth object detec tion labeling task. Use Amazon Mechanical", + "C. Use Amazon Rekognition Custom Labels to label the dataset and create a custom Amazon", + "D. Use an Amazon SageMaker Ground Truth semantic seg mentation labeling task. Use a private" + ], + "correct": "D. Use an Amazon SageMaker Ground Truth semantic seg mentation labeling task. Use a private", + "explanation": "", + "references": "" + }, + { + "question": "A data engineer at a bank is evaluating a new tabul ar dataset that includes customer dat", + "options": [ + "A. The data engineer will use the customer data to c reate a new model to predict customer behavior.", + "B. Use a linear-based algorithm to train the model.", + "C. Apply principal component analysis (PCA).", + "D. Remove a portion of highly correlated features fr om the dataset." + ], + "correct": "", + "explanation": "B) Apply principal component analysis (PCA): PCA is a technique that reduces the dimensionality of a dataset by transforming the original features into a smaller set of new features that capture most of the variance in the data. PCA can help address the issue of multicollinearity, which occurs when some features are highly correlated with each other and can cause problems for some machine learning algorithms. By applying PCA, the data engi neer can reduce the number of features and remove the redundancy in the data. C) Remove a portion of highly correlated features f rom the dataset: Another way to deal with multicollinearity is to manually remove some of the features that are highly correlated with each other. This can help simplify the model and avoid o verfitting. The data engineer can use the correlation matrix to identify the features that ha ve a high correlation coefficient (e.g., above 0.8 or below -0.8) and remove one of them from the dataset . Reference: = Principal Component Analysis: This is a document fr om AWS that explains what PCA is, how it works, and how to use it with Amazon SageMaker. Multicollinearity: This is a document from AWS that describes what multicollinearity is, how to detect it, and how to deal with it.", + "references": "" + }, + { + "question": "A company is building a new version of a recommenda tion engine. Machine learning (ML) specialists need to keep adding new data from users to improve personalized recommendations. The ML specialists gather data from the users interactions on the platform and from sources such as external websites and social media. The pipeline cleans, transforms, enriches, and comp resses terabytes of data daily, and this data is stored in Amazon S3. A set of Python scripts was co ded to do the job and is stored in a large Amazon EC2 instance. The whole process takes more than 20 hours to finish, with each script taking at least an hour. The company wants to move the scripts out of Amazon EC2 into a more managed solution that will eliminate the need to maintain servers. Which approach will address all of these requiremen ts with the LEAST development effort?", + "options": [ + "A. Load the data into an Amazon Redshift cluster. Ex ecute the pipeline by using SQL. Store the results", + "B. Load the data into Amazon DynamoDB. Convert the s cripts to an AWS Lambda function. Execute", + "C. Create an AWS Glue job. Convert the scripts to Py Spark. Execute the pipeline. Store the results in", + "D. Create a set of individual AWS Lambda functions t o execute each of the scripts. Build a step function by using the AWS Step Functions Data Scien ce SDK. Store the results in Amazon S3." + ], + "correct": "C. Create an AWS Glue job. Convert the scripts to Py Spark. Execute the pipeline. Store the results in", + "explanation": "The best approach to address all of the requirement s with the least development effort is to create an AWS Glue job, convert the scripts to PySpark, ex ecute the pipeline, and store the results in Amazon S3. This is because: AWS Glue is a fully managed extract, transform, and load (ETL) service that makes it easy to prepare and load data for analytics 1. AWS Glue can run Pyt hon and Scala scripts to process data from various sources, such as Amazon S3, Amazon DynamoDB, Amazon Redshift, and more 2. AWS Glue also provides a serverless Apache Spark environment to r un ETL jobs, eliminating the need to provision and manage servers 3. PySpark is the Python API for Apache Spark, a unifi ed analytics engine for large-scale data processing 4. PySpark can perform various data tran sformations and manipulations on structured and unstructured data, such as cleaning, enriching, and compressing 5. PySpark can also leverage the distributed computing power of Spark to handle tera bytes of data efficiently and scalably 6. By creating an AWS Glue job and converting the scri pts to PySpark, the company can move the scripts out of Amazon EC2 into a more managed solut ion that will eliminate the need to maintain servers. The company can also reduce the developmen t effort by using the AWS Glue console, AWS SDK, or AWS CLI to create and run the job 7. Moreov er, the company can use the AWS Glue Data Catalog to store and manage the metadata of the dat a sources and targets 8. The other options are not as suitable as option C f or the following reasons: Option A is not optimal because loading the data in to an Amazon Redshift cluster and executing the pipeline by using SQL will incur additional costs a nd complexity for the company. Amazon Redshift is a fully managed data warehouse service that enables fast and scalable analysis of structured data . However, it is not designed for ETL purposes, such as cleaning, transforming, enriching, and compressing data. Moreover, using SQL to perform th ese tasks may not be as expressive and flexible as using Python scripts. Furthermore, the company w ill have to provision and configure the Amazon Redshift cluster, and load and unload the data from Amazon S3, which will increase the development effort and time. Option B is not feasible because loading the data i nto Amazon DynamoDB and converting the scripts to an AWS Lambda function will not work for the com panys use case. Amazon DynamoDB is a fully managed key-value and document database service tha t provides fast and consistent performance at any scale . However, it is not suitable for storing and processing terabytes of data daily, as it has limits on the size and throughput of each table and item . Moreover, using AWS Lambda to execute the pipeline will not be efficient or cost-effective, a s Lambda has limits on the memory, CPU, and execution time of each function . Therefore, using Amazon DynamoDB and AWS Lambda will not meet the companys requirements for processing large amounts of data quickly and reliably. Option D is not relevant because creating a set of individual AWS Lambda functions to execute each of the scripts and building a step function by usin g the AWS Step Functions Data Science SDK will not address the main issue of moving the scripts out of Amazon EC2. AWS Step Functions is a fully managed service that lets you coordinate multiple A WS services into serverless workflows . The AWS Step Functions Data Science SDK is an open source l ibrary that allows data scientists to easily create workflows that process and publish machine learning models using Amazon SageMaker and AWS Step Functions . However, these services and tools are not designed for ETL purposes, such as cleaning, transforming, enriching, and compressing data. Moreover, as mentioned in option B, using AWS Lambda to execute the scripts will not be effic ient or cost-effective for the companys use case.", + "references": "What Is AWS Glue? AWS Glue Components AWS Glue Serverless Spark ETL PySpark - Overview PySpark - RDD PySpark - SparkContext Adding Jobs in AWS Glue Populating the AWS Glue Data Catalog [What Is Amazon Redshift?] [What Is Amazon DynamoDB?] [Service, Account, and Table Quotas in DynamoDB] [AWS Lambda quotas] [What Is AWS Step Functions?] [AWS Step Functions Data Science SDK for Python]" + }, + { + "question": "A retail company is selling products through a glob al online marketplace. The company wants to use machine learning (ML) to analyze customer feedback and identify specific areas for improvement. A developer has built a tool that collects customer r eviews from the online marketplace and stores them in an Amazon S3 bucket. This process yields a dataset of 40 reviews. A data scientist building the ML models must identify additional sources of d ata to increase the size of the dataset. Which data sources should the data scientist use to augment the dataset of reviews? (Choose three.)", + "options": [ + "A. Emails exchanged by customers and the companys cu stomer service agents", + "B. Social media posts containing the name of the com pany or its products", + "C. A publicly available collection of news articles", + "D. A publicly available collection of customer revie ws" + ], + "correct": "", + "explanation": "The data sources that the data scientist should use to augment the dataset of reviews are those that contain relevant and diverse customer feedback abou t the company or its products. Emails exchanged by customers and the companys customer se rvice agents can provide valuable insights into the issues and complaints that customers have, as well as the solutions and responses that the company offers. Social media posts containing the n ame of the company or its products can capture the opinions and sentiments of customers and potent ial customers, as well as their reactions to marketing campaigns and product launches. A publicl y available collection of customer reviews can provide a large and varied sample of feedback from different online platforms and marketplaces, which can help to generalize the ML models and avoi d bias.", + "references": "Detect sentiment from customer reviews using Amazon Comprehend | AWS Machine Learning Blog How to Apply Machine Learning to Customer Feedback" + }, + { + "question": "A machine learning (ML) specialist wants to create a data preparation job that uses a PySpark script with complex window aggregation operations to creat e data for training and testing. The ML specialist needs to evaluate the impact of the numb er of features and the sample count on model performance. Which approach should the ML specialist use to dete rmine the ideal data transformations for the model?", + "options": [ + "A. Add an Amazon SageMaker Debugger hook to the scri pt to capture key metrics. Run the script as", + "B. Add an Amazon SageMaker Experiments tracker to th e script to capture key metrics. Run the script", + "C. Add an Amazon SageMaker Debugger hook to the scri pt to capture key parameters. Run the script", + "D. Add an Amazon SageMaker Experiments tracker to th e script to capture key parameters. Run the" + ], + "correct": "", + "explanation": "Amazon SageMaker Experiments is a service that help s track, compare, and evaluate different iterations of ML models. It can be used to capture key parameters such as the number of features and the sample count from a PySpark script that run s as a SageMaker processing job. A SageMaker processing job is a flexible and scalable way to ru n data processing workloads on AWS, such as feature engineering, data validation, model evaluat ion, and model interpretation.", + "references": "Amazon SageMaker Experiments Process Data and Evaluate Models" + }, + { + "question": "A data scientist has a dataset of machine part imag es stored in Amazon Elastic File System (Amazon EFS). The data scientist needs to use Amazon SageMa ker to create and train an image classification machine learning model based on this dataset. Becau se of budget and time constraints, management wants the data scientist to create and t rain a model with the least number of steps and integration work required. How should the data scientist meet these requiremen ts?", + "options": [ + "A. Mount the EFS file system to a SageMaker notebook and run a script that copies the data to an", + "B. Launch a transient Amazon EMR cluster. Configure steps to mount the EFS file system and copy the", + "C. Mount the EFS file system to an Amazon EC2 instan ce and use the AWS CLI to copy the data to an", + "D. Run a SageMaker training job with an EFS file sys tem as the data source." + ], + "correct": "D. Run a SageMaker training job with an EFS file sys tem as the data source.", + "explanation": "The simplest and fastest way to use the EFS dataset for SageMaker training is to run a SageMaker training job with an EFS file system as the data so urce. This option does not require any data copying or additional integration steps. SageMaker supports EFS as a data source for training jobs, and it can mount the EFS file system to the training container using the FileSystemConfig parameter. This way, the training script can access the data files as if they were on the local disk of the training instance. Reference: Access Training Data - Amazon SageMaker Mount an EFS file system to an Amazon SageMaker not ebook (with lifecycle configurations) | AWS Machine Learning Blog", + "references": "" + }, + { + "question": "A retail company uses a machine learning (ML) model for daily sales forecasting. The companys brand manager reports that the model has provided i naccurate results for the past 3 weeks. At the end of each day, an AWS Glue job consolidate s the input data that is used for the forecasting with the actual daily sales data and the prediction s of the model. The AWS Glue job stores the data in Amazon S3. The companys ML team is using an Amazon SageMaker Studio notebook to gain an understanding about the source of the model's inacc uracies. What should the ML team do on the SageMaker Studio notebook to visualize the model's degradation MOST accurately? A. Create a histogram of the daily sales over the la st 3 weeks. In addition, create a histogram of the daily sales from before that period.", + "options": [ + "B. Create a histogram of the model errors over the l ast 3 weeks. In addition, create a histogram of", + "C. Create a line chart with the weekly mean absolute error (MAE) of the model.", + "D. Create a scatter plot of daily sales versus model error for the last 3 weeks. In addition, create a" + ], + "correct": "B. Create a histogram of the model errors over the l ast 3 weeks. In addition, create a histogram of", + "explanation": "The best way to visualize the models degradation is to create a histogram of the model errors over the last 3 weeks and compare it with a histogram of the model errors from before that period. A histogram is a graphical representation of the dist ribution of numerical data. It shows how often each value or range of values occurs in the data. A mode l error is the difference between the actual value and the predicted value. A high model error indicat es a poor fit of the model to the data. By comparing the histograms of the model errors, the M L team can see if there is a significant change in the shape, spread, or center of the distribution. T his can indicate if the model is underfitting, overfitting, or drifting from the data. A line char t or a scatter plot would not be as effective as a histogram for this purpose, because they do not sho w the distribution of the errors. A line chart would only show the trend of the errors over time, which may not capture the variability or outliers. A scatter plot would only show the relationship bet ween the errors and another variable, such as daily sales, which may not be relevant or informati ve for the models performance. Reference: Histogram - Wikipedia Model error - Wikipedia SageMaker Model Monitor - visualizing monitoring re sults", + "references": "" + }, + { + "question": "An ecommerce company sends a weekly email newslette r to all of its customers. Management has hired a team of writers to create additional target ed content. A data scientist needs to identify five customer segments based on age, income, and locatio n. The customers current segmentation is unknown. The data scientist previously built an XGB oost model to predict the likelihood of a customer responding to an email based on age, incom e, and location. Why does the XGBoost model NOT meet the current req uirements, and how can this be fixed?", + "options": [ + "A. The XGBoost model provides a true/false binary ou tput. Apply principal component analysis (PCA)", + "B. The XGBoost model provides a true/false binary ou tput. Increase the number of classes the", + "C. The XGBoost model is a supervised machine learnin g algorithm. Train a k-Nearest-Neighbors (kNN)", + "D. The XGBoost model is a supervised machine learnin g algorithm. Train a k-means model with K = 5" + ], + "correct": "D. The XGBoost model is a supervised machine learnin g algorithm. Train a k-means model with K = 5", + "explanation": "The XGBoost model is a supervised machine learning algorithm, which means it requires labeled data to learn from. The customers current segmentat ion is unknown, so there is no label to train the XGBoost model on. Moreover, the XGBoost model is de signed for classification or regression tasks, not for clustering. Clustering is a type of unsuper vised machine learning, which means it does not require labeled data. Clustering algorithms try to find natural groups or clusters in the data based o n their similarity or distance. A common clustering a lgorithm is k-means, which partitions the data into K clusters, where each data point belongs to the cl uster with the nearest mean. To meet the current requirements, the data scientist should train a k-m eans model with K = 5 on the same dataset to predict a segment for each customer. This way, the data scientist can identify five customer segments based on age, income, and location, without needing any labels. Reference: What is XGBoost? - Amazon SageMaker What is Clustering? - Amazon SageMaker K-Means Algorithm - Amazon SageMaker", + "references": "" + }, + { + "question": "A global financial company is using machine learnin g to automate its loan approval process. The company has a dataset of customer information. The dataset contains some categorical fields, such as customer location by city and housing status. Th e dataset also includes financial fields in differe nt units, such as account balances in US dollars and m onthly interest in US cents. The companys data scientists are using a gradient b oosting regression model to infer the credit score for each customer. The model has a training accurac y of 99% and a testing accuracy of 75%. The data scientists want to improve the models testing accur acy. Which process will improve the testing accuracy the MOST?", + "options": [ + "A. Use a one-hot encoder for the categorical fields in the dataset. Perform standardization on the", + "B. Use tokenization of the categorical fields in the dataset. Perform binning on the financial fields i n", + "C. Use a label encoder for the categorical fields in the dataset. Perform L1 regularization on the", + "D. Use a logarithm transformation on the categorical fields in the dataset. Perform binning on the" + ], + "correct": "A. Use a one-hot encoder for the categorical fields in the dataset. Perform standardization on the", + "explanation": "The question is about improving the testing accurac y of a gradient boosting regression model. The testing accuracy is much lower than the training ac curacy, which indicates that the model is overfitting the training data. To reduce overfittin g, the following steps are recommended: Use a one-hot encoder for the categorical fields in the dataset. This will create binary features for each category and avoid imposing an ordinal relatio nship among them. This can help the model learn the patterns better and generalize to unseen data. Perform standardization on the financial fields in the dataset. This will scale the features to have z ero mean and unit variance, which can improve the conve rgence and performance of the model. This can also help the model handle features with different units and ranges. Apply L1 regularization to the data. This will add a penalty term to the loss function that is proportional to the absolute value of the coefficie nts. This can help the model reduce the complexity and select the most relevant features by shrinking the coefficients of less important features to zero .", + "references": "1: AWS Machine Learning Specialty Exam Guide 2: AWS Machine Learning Specialty Course 3: AWS Machine Learning Blog" + }, + { + "question": "A machine learning (ML) specialist needs to extract embedding vectors from a text series. The goal is to provide a ready-to-ingest feature space for a da ta scientist to develop downstream ML predictive models. The text consists of curated sentences in E nglish. Many sentences use similar words but in different contexts. There are questions and answers among the sentences, and the embedding space must differentiate between them. Which options can produce the required embedding ve ctors that capture word context and sequential QA information? (Choose two.)", + "options": [ + "A. Amazon SageMaker seq2seq algorithm", + "B. Amazon SageMaker BlazingText algorithm in Skip-gr am mode", + "C. Amazon SageMaker Object2Vec algorithm", + "D. Amazon SageMaker BlazingText algorithm in continu ous bag-of-words (CBOW) mode" + ], + "correct": "", + "explanation": "To capture word context and sequential QA informati on, the embedding vectors need to consider both the order and the meaning of the words in the text. Option B, Amazon SageMaker BlazingText algorithm in Skip-gram mode, is a valid option because it can learn word embeddings that capture the semantic similarity and syntactic relations between words based on their co-occurrence in a window of w ords. Skip-gram mode can also handle rare words better than continuous bag-of-words (CBOW) mo de1. Option E, combination of the Amazon SageMaker Blazi ngText algorithm in Batch Skip-gram mode with a custom recurrent neural network (RNN), is an other valid option because it can leverage the advantages of Skip-gram mode and also use an RNN to model the sequential nature of the text. An RNN can capture the temporal dependencies and long- term dependencies between words, which are important for QA tasks2. Option A, Amazon SageMaker seq2seq algorithm, is no t a valid option because it is designed for sequence-to-sequence tasks such as machine translat ion, summarization, or chatbots. It does not produce embedding vectors for text series, but rath er generates an output sequence given an input sequence3. Option C, Amazon SageMaker Object2Vec algorithm, is not a valid option because it is designed for learning embeddings for pairs of objects, such as t ext-image, text-text, or image-image. It does not produce embedding vectors for text series, but rath er learns a similarity function between pairs of objects4. Option D, Amazon SageMaker BlazingText algorithm in continuous bag-of-words (CBOW) mode, is not a valid option because it does not capture word context as well as Skip-gram mode. CBOW mode predicts a word given its surrounding words, while Skip-gram mode predicts the surrounding words given a word. CBOW mode is faster and more suitable for frequent words, but Skip-gram mode can learn more meaningful embeddings for rare words1.", + "references": "1: Amazon SageMaker BlazingText 2: Recurrent Neural Networks (RNNs) 3: Amazon SageMaker Seq2Seq 4: Amazon SageMaker Object2Vec" + }, + { + "question": "A retail company wants to update its customer suppo rt system. The company wants to implement automatic routing of customer claims to different q ueues to prioritize the claims by category. Currently, an operator manually performs the catego ry assignment and routing. After the operator classifies and routes the claim, the company stores the claims record in a central database. The claims record includes the claims category. The company has no data science team or experience in the field of machine learning (ML). The companys small development team needs a solution th at requires no ML expertise. Which solution meets these requirements?", + "options": [ + "A. Export the database to a .csv file with two colum ns: claim_label and claim_text. Use the Amazon", + "B. Export the database to a .csv file with one colum n: claim_text. Use the Amazon SageMaker Latent", + "C. Use Amazon Textract to process the database and a utomatically detect two columns: claim_label", + "D. Export the database to a .csv file with two colum ns: claim_label and claim_text. Use Amazon" + ], + "correct": "D. Export the database to a .csv file with two colum ns: claim_label and claim_text. Use Amazon", + "explanation": "Amazon Comprehend is a natural language processing (NLP) service that can analyze text and extract insights such as sentiment, entities, topics, and l anguage. Amazon Comprehend also provides custom classification and custom entity recognition features that allow users to train their own models using their own data and labels. For the sce nario of routing customer claims to different queues based on categories, Amazon Comprehend custo m classification is a suitable solution. The custom classifier can be trained using a .csv file that contains the claim text and the claim label as columns. The custom classifier can then be used to process incoming claims and predict the labels using the Amazon Comprehend API. The predicted labe ls can be used to route the claims to the appropriate queue. This solution does not require a ny machine learning expertise or model deployment, and it can be easily integrated with th e existing application. The other options are not suitable because: Option A: Amazon SageMaker Object2Vec is an algorit hm that can learn embeddings of objects such as words, sentences, or documents. It can be used f or tasks such as text classification, sentiment analysis, or recommendation systems. However, using this algorithm requires machine learning expertise and model deployment using SageMaker, whi ch are not available for the company. Option B: Amazon SageMaker Latent Dirichlet Allocat ion (LDA) is an algorithm that can discover the topics or themes in a collection of documents. It c an be used for tasks such as topic modeling, document clustering, or text summarization. However , using this algorithm requires machine learning expertise and model deployment using SageM aker, which are not available for the company. Moreover, LDA does not provide labels for the topic s, but rather a distribution of words for each topic, which may not match the existing categories of the claims. Option C: Amazon Textract is a service that can ext ract text and data from scanned documents or images. It can be used for tasks such as document a nalysis, data extraction, or form processing. However, using this service is unnecessary and inef ficient for the scenario, since the company already has the claim text and label in a database. Moreover, Amazon Textract does not provide custom classification features, so it cannot be use d to train a custom classifier using the existing d ata and labels.", + "references": "Amazon Comprehend Custom Classification Amazon SageMaker Object2Vec Amazon SageMaker Latent Dirichlet Allocation Amazon Textract" + }, + { + "question": "A machine learning (ML) specialist is using Amazon SageMaker hyperparameter optimization (HPO) to improve a models accuracy. The learning rate par ameter is specified in the following HPO configuration: During the results analysis, the ML specialist dete rmines that most of the training jobs had a learnin g rate between 0.01 and 0.1. The best result had a le arning rate of less than 0.01. Training jobs need t o run regularly over a changing dataset. The ML speci alist needs to find a tuning mechanism that uses different learning rates more evenly from the provi ded range between MinValue and MaxValue. Which solution provides the MOST accurate result?", + "options": [ + "A. Modify the HPO configuration as follows:", + "B. Run three different HPO jobs that use different l earning rates form the following intervals for", + "C. Modify the HPO configuration as follows:", + "D. Run three different HPO jobs that use different l earning rates form the following intervals for" + ], + "correct": "", + "explanation": "The solution C modifies the HPO configuration to us e a logarithmic scale for the learning rate parameter. This means that the values of the learni ng rate are sampled from a log-uniform distribution, which gives more weight to smaller va lues. This can help to explore the lower end of the range more evenly and find the optimal learning rat e more efficiently. The other solutions either use a linear scale, which may not sample enough values from the lower end, or divide the range into subint ervals, which may miss some combinations of hyperparameters . Reference: How Hyperparameter Tuning Works - Amazon SageMaker Tuning Hyperparameters - Amazon SageMaker", + "references": "" + }, + { + "question": "A manufacturing company wants to use machine learni ng (ML) to automate quality control in its facilities. The facilities are in remote locations and have limited internet connectivity. The company has 20 \u00d0\u00a2\u00d0' of training data that consists of label ed images of defective product parts. The training data is in the corporate on-premises data center. The company will use this data to train a model for real-time defect detection in new parts as the parts move on a conveyor belt in the facilities. Th e company needs a solution that minimizes costs for compute infrastructure and that maximizes the s calability of resources for training. The solution also must facilitate the companys use of an ML mode l in the low-connectivity environments. Which solution will meet these requirements?", + "options": [ + "A. Move the training data to an Amazon S3 bucket. Tr ain and evaluate the model by using Amazon", + "B. Train and evaluate the model on premises. Upload the model to an Amazon S3 bucket. Deploy the", + "C. Move the training data to an Amazon S3 bucket. Tr ain and evaluate the model by using Amazon", + "D. Train the model on premises. Upload the model to an Amazon S3 bucket. Set up an edge device in" + ], + "correct": "C. Move the training data to an Amazon S3 bucket. Tr ain and evaluate the model by using Amazon", + "explanation": "The solution C meets the requirements because it mi nimizes costs for compute infrastructure, maximizes the scalability of resources for training , and facilitates the use of an ML model in lowconn ectivity environments. The solution C involves the following steps: Move the training data to an Amazon S3 bucket. This will enable the company to store the large amount of data in a durable, scalable, and cost-eff ective way. It will also allow the company to acces s the data from the cloud for training and evaluation purposes1. Train and evaluate the model by using Amazon SageMa ker. This will enable the company to use a fully managed service that provides various feature s and tools for building, training, tuning, and deploying ML models. Amazon SageMaker can handle la rge-scale data processing and distributed training, and it can leverage the power of AWS comp ute resources such as Amazon EC2, Amazon EKS, and AWS Fargate2. Optimize the model by using SageMaker Neo. This wil l enable the company to reduce the size of the model and improve its performance and efficiency. S ageMaker Neo can compile the model into an executable that can run on various hardware platfor ms, such as CPUs, GPUs, and edge devices3. Set up an edge device in the manufacturing faciliti es with AWS IoT Greengrass. This will enable the company to deploy the model on a local device that can run inference in real time, even in lowconnecti vity environments. AWS IoT Greengrass can extend AWS clo ud capabilities to the edge, and it can securely communicate with the cloud for upda tes and synchronization4. Deploy the model on the edge device. This will enab le the company to automate quality control in its facilities by using the model to detect defects in new parts as they move on a conveyor belt. The model can run inference locally on the edge device without requiring internet connectivity, and it can send the results to the cloud when the connection i s available4. The other options are not suitable because: Option A: Deploying the model on a SageMaker hostin g services endpoint will not facilitate the use of the model in low-connectivity environments, as i t will require internet access to perform inference. Moreover, it may incur higher costs for hosting and data transfer than deploying the model on an edge device. Option B: Training and evaluating the model on prem ises will not minimize costs for compute infrastructure, as it will require the company to m aintain and upgrade its own hardware and software. Moreover, it will not maximize the scalab ility of resources for training, as it will limit t he companys ability to leverage the clouds elasticity and flexibility. Option D: Training the model on premises will not m inimize costs for compute infrastructure, nor maximize the scalability of resources for training, for the same reasons as option B.", + "references": "1: Amazon S3 2: Amazon SageMaker 3: SageMaker Neo 4: AWS IoT Greengrass" + }, + { + "question": "A company has an ecommerce website with a product r ecommendation engine built in TensorFlow. The recommendation engine endpoint is hosted by Ama zon SageMaker. Three compute-optimized instances support the expected peak load of the web site. Response times on the product recommendation page a re increasing at the beginning of each month. Some users are encountering errors. The webs ite receives the majority of its traffic between 8 AM and 6 PM on weekdays in a single time zone. Which of the following options are the MOST effecti ve in solving the issue while keeping costs to a minimum? (Choose two.)", + "options": [ + "A. Configure the endpoint to use Amazon Elastic Infe rence (EI) accelerators.", + "B. Create a new endpoint configuration with two prod uction variants.", + "C. Configure the endpoint to automatically scale wit h the Invocations Per Instance metric.", + "D. Deploy a second instance pool to support a blue/g reen deployment of models." + ], + "correct": "", + "explanation": "The solution A and C are the most effective in solv ing the issue while keeping costs to a minimum. The solution A and C involve the following steps: Configure the endpoint to use Amazon Elastic Infere nce (EI) accelerators. This will enable the company to reduce the cost and latency of running T ensorFlow inference on SageMaker. Amazon EI provides GPU-powered acceleration for deep learning models without requiring the use of GPU instances. Amazon EI can attach to any SageMaker in stance type and provide the right amount of acceleration based on the workload1. Configure the endpoint to automatically scale with the Invocations Per Instance metric. This will enable the company to adjust the number of instance s based on the demand and traffic patterns of the website. The Invocations Per Instance metric me asures the average number of requests that each instance processes over a period of time. By using this metric, the company can scale out the endpoint when the load increases and scale in when the load decreases. This can improve the response time and availability of the product recom mendation engine2. The other options are not suitable because: Option B: Creating a new endpoint configuration wit h two production variants will not solve the issue of increasing response time and errors. Production variants are used to split the traffic between different models or versions of the same model. The y can be useful for testing, updating, or A/B testing models. However, they do not provide any sc aling or acceleration benefits for the inference workload3. Option D: Deploying a second instance pool to suppo rt a blue/green deployment of models will not solve the issue of increasing response time and err ors. Blue/green deployment is a technique for updating models without downtime or disruption. It involves creating a new endpoint configuration with a different instance pool and model version, a nd then shifting the traffic from the old endpoint to the new endpoint gradually. However, this techni que does not provide any scaling or acceleration benefits for the inference workload4. Option E: Reconfiguring the endpoint to use burstab le instances will not solve the issue of increasing response time and errors. Burstable instances are i nstances that provide a baseline level of CPU performance with the ability to burst above the bas eline when needed. They can be useful for workloads that have moderate CPU utilization and oc casional spikes. However, they are not suitable for workloads that have high and consistent CPU uti lization, such as the product recommendation engine. Moreover, burstable instances may incur add itional charges when they exceed their CPU credits5.", + "references": "1: Amazon Elastic Inference 2: How to Scale Amazon SageMaker Endpoints 3: Deploying Models to Amazon SageMaker Hosting Ser vices 4: Updating Models in Amazon SageMaker Hosting Serv ices 5: Burstable Performance Instances" + }, + { + "question": "A real-estate company is launching a new product th at predicts the prices of new houses. The historical data for the properties and prices is st ored in .csv format in an Amazon S3 bucket. The dat a has a header, some categorical fields, and some mis sing values. The companys data scientists have used Python with a common open-source library to fi ll the missing values with zeros. The data scientists have dropped all of the categorical fiel ds and have trained a model by using the opensource linear regression algorithm with the default parame ters. The accuracy of the predictions with the current mo del is below 50%. The company wants to improve the model performance and launch the new product as soon as possible. Which solution will meet these requirements with th e LEAST operational overhead?", + "options": [ + "A. Create a service-linked role for Amazon Elastic C ontainer Service (Amazon ECS) with access to the", + "B. Create an Amazon SageMaker notebook with a new IA M role that is associated with the notebook.", + "C. Create an IAM role with access to Amazon S3, Amaz on SageMaker, and AWS Lambda. Create a", + "D. Create an IAM role for Amazon SageMaker with acce ss to the S3 bucket. Create a SageMaker" + ], + "correct": "D. Create an IAM role for Amazon SageMaker with acce ss to the S3 bucket. Create a SageMaker", + "explanation": "The solution C is the best option to identify and a ddress training issues with the least development effort. The solution C involves the following steps : Use the SageMaker Debugger vanishing_gradient and L owGPUUtilization built-in rules to detect issues. SageMaker Debugger is a feature of Amazon S ageMaker that allows data scientists to monitor, analyze, and debug machine learning models during training. SageMaker Debugger provides a set of built-in rules that can automatic ally detect common issues and anomalies in model training, such as vanishing or exploding gradients, overfitting, underfitting, low GPU utilization, an d more1. The data scientist can use the vanishing_gra dient rule to check if the gradients are becoming too small and causing the training to not converge. The data scientist can also use the LowGPUUtilization rule to check if the GPU resource s are underutilized and causing the training to be inefficient2. Launch the StopTrainingJob action if issues are det ected. SageMaker Debugger can also take actions based on the status of the rules. One of the action s is StopTrainingJob, which can terminate the training job if a rule is in an error state. This c an help the data scientist to save time and money b y stopping the training early if issues are detected3 . The other options are not suitable because: Option A: Using CPU utilization metrics that are ca ptured in Amazon CloudWatch and configuring a CloudWatch alarm to stop the training job early if low CPU utilization occurs will not identify and address training issues effectively. CPU utilizatio n is not a good indicator of model training performance, especially for GPU instances. Moreover , CloudWatch alarms can only trigger actions based on simple thresholds, not complex rules or co nditions4. Option B: Using high-resolution custom metrics that are captured in Amazon CloudWatch and configuring an AWS Lambda function to analyze the m etrics and to stop the training job early if issues are detected will incur more development effort tha n using SageMaker Debugger. The data scientist will have to write the code for capturing, sending, and analyzing the custom metrics, as well as for invoking the Lambda function and stopping the train ing job. Moreover, this solution may not be able to detect all the issues that SageMaker Debugger ca n5. Option D: Using the SageMaker Debugger confusion an d feature_importance_overweight built-in rules and launching the StopTrainingJob action if i ssues are detected will not identify and address training issues effectively. The confusion rule is used to monitor the confusion matrix of a classification model, which is not relevant for a r egression model that predicts prices. The feature_importance_overweight rule is used to check if some features have too much weight in the model, which may not be related to the convergence or resource utilization issues2.", + "references": "1: Amazon SageMaker Debugger 2: Built-in Rules for Amazon SageMaker Debugger 3: Actions for Amazon SageMaker Debugger 4: Amazon CloudWatch Alarms 5: Amazon CloudWatch Custom Metrics" + }, + { + "question": "A company needs to deploy a chatbot to answer commo n questions from customers. The chatbot must base its answers on company documentation. Which solution will meet these requirements with th e LEAST development effort?", + "options": [ + "A. Index company documents by using Amazon Kendra. I ntegrate the chatbot with Amazon Kendra", + "B. Train a Bidirectional Attention Flow (BiDAF) netw ork based on past customer questions and", + "C. Train an Amazon SageMaker BlazingText model based on past customer questions and company", + "D. Index company documents by using Amazon OpenSearc h Service. Integrate the chatbot with" + ], + "correct": "A. Index company documents by using Amazon Kendra. I ntegrate the chatbot with Amazon Kendra", + "explanation": "The solution A will meet the requirements with the least development effort because it uses Amazon Kendra, which is a highly accurate and easy to use intelligent search service powered by machine learning. Amazon Kendra can index company documents from various sources and formats, such as PDF, HTML, Word, and more. Amazon Kendra can also i ntegrate with chatbots by using the Amazon Kendra Query API operation, which can understand na tural language questions and provide relevant answers from the indexed documents. Amazon Kendra c an also provide additional information, such as document excerpts, links, and FAQs, to enhance t he chatbot experience1. The other options are not suitable because: Option B: Training a Bidirectional Attention Flow ( BiDAF) network based on past customer questions and company documents, deploying the model as a rea l-time Amazon SageMaker endpoint, and integrating the model with the chatbot by using the SageMaker Runtime InvokeEndpoint API operation will incur more development effort than u sing Amazon Kendra. The company will have to write the code for the BiDAF network, which is a co mplex deep learning model for question answering. The company will also have to manage the SageMaker endpoint, the model artifact, and the inference logic2. Option C: Training an Amazon SageMaker BlazingText model based on past customer questions and company documents, deploying the model as a real-ti me SageMaker endpoint, and integrating the model with the chatbot by using the SageMaker Runti me InvokeEndpoint API operation will incur more development effort than using Amazon Kendra. T he company will have to write the code for the BlazingText model, which is a fast and scalable text classification and word embedding algorithm. The company will also have to manage the SageMaker endpoint, the model artifact, and the inference logic3. Option D: Indexing company documents by using Amazo n OpenSearch Service and integrating the chatbot with OpenSearch Service by using the OpenSe arch Service k-nearest neighbors (k-NN) Query API operation will not meet the requirements effect ively. Amazon OpenSearch Service is a fully managed service that provides fast and scalable sea rch and analytics capabilities. However, it is not designed for natural language question answering, a nd it may not provide accurate or relevant answers for the chatbot. Moreover, the k-NN Query A PI operation is used to find the most similar documents or vectors based on a distance function, not to find the best answers based on a natural language query4.", + "references": "1: Amazon Kendra 2: Bidirectional Attention Flow for Machine Compreh ension 3: Amazon SageMaker BlazingText 4: Amazon OpenSearch Service" + }, + { + "question": "A company ingests machine learning (ML) data from w eb advertising clicks into an Amazon S3 data lake. Click data is added to an Amazon Kinesis data stream by using the Kinesis Producer Library (KPL). The data is loaded into the S3 data lake fro m the data stream by using an Amazon Kinesis Data Firehose delivery stream. As the data volume increa ses, an ML specialist notices that the rate of data ingested into Amazon S3 is relatively constant. The re also is an increasing backlog of data for Kinesi s Data Streams and Kinesis Data Firehose to ingest. Which next step is MOST likely to improve the data ingestion rate into Amazon S3?", + "options": [ + "A. Increase the number of S3 prefixes for the delive ry stream to write to.", + "B. Decrease the retention period for the data stream .", + "C. Increase the number of shards for the data stream .", + "D. Add more consumers using the Kinesis Client Libra ry (KCL)." + ], + "correct": "C. Increase the number of shards for the data stream .", + "explanation": "The solution C is the most likely to improve the da ta ingestion rate into Amazon S3 because it increases the number of shards for the data stream. The number of shards determines the throughput capacity of the data stream, which affec ts the rate of data ingestion. Each shard can support up to 1 MB per second of data input and 2 M B per second of data output. By increasing the number of shards, the company can increase the data ingestion rate proportionally. The company can use the UpdateShardCount API operation to modify th e number of shards in the data stream1. The other options are not likely to improve the dat a ingestion rate into Amazon S3 because: Option A: Increasing the number of S3 prefixes for the delivery stream to write to will not affect the data ingestion rate, as it only changes the way the data is organized in the S3 bucket. The number of S3 prefixes can help to optimize the performance of downstream applications that read the data from S3, but it does not impact the performance of Kinesis Data Firehose2. Option B: Decreasing the retention period for the d ata stream will not affect the data ingestion rate, as it only changes the amount of time the data is s tored in the data stream. The retention period can help to manage the data availability and durability , but it does not impact the throughput capacity of the data stream3. Option D: Adding more consumers using the Kinesis C lient Library (KCL) will not affect the data ingestion rate, as it only changes the way the data is processed by downstream applications. The consumers can help to scale the data processing and handle failures, but they do not impact the data ingestion into S3 by Kinesis Data Firehose4.", + "references": "1: Resharding - Amazon Kinesis Data Streams 2: Amazon S3 Prefixes - Amazon Kinesis Data Firehos e 3: Data Retention - Amazon Kinesis Data Streams 4: Developing Consumers Using the Kinesis Client Li brary - Amazon Kinesis Data Streams" + }, + { + "question": "A manufacturing company has a production line with sensors that collect hundreds of quality metrics. The company has stored sensor data and man ual inspection results in a data lake for several months. To automate quality control, the machine le arning team must build an automated mechanism that determines whether the produced good s are good quality, replacement market quality, or scrap quality based on the manual inspe ction results. Which modeling approach will deliver the MOST accur ate prediction of product quality?", + "options": [ + "A. Amazon SageMaker DeepAR forecasting algorithm", + "B. Amazon SageMaker XGBoost algorithm", + "C. Amazon SageMaker Latent Dirichlet Allocation (LDA ) algorithm", + "D. A convolutional neural network (CNN) and ResNet" + ], + "correct": "D. A convolutional neural network (CNN) and ResNet", + "explanation": "A convolutional neural network (CNN) is a type of d eep learning model that can learn to extract features from images and perform tasks such as clas sification, segmentation, and detection1. ResNet is a popular CNN architecture that uses residual co nnections to overcome the problem of vanishing gradients and enable very deep networks2. For the t ask of predicting product quality based on sensor data, a CNN and ResNet approach can leverage the spatial structure of the data and learn complex patterns that distinguish different quality levels.", + "references": "Convolutional Neural Networks (CNNs / ConvNets) PyTorch ResNet: The Basics and a Quick Tutorial" + }, + { + "question": "A media company wants to create a solution that ide ntifies celebrities in pictures that users upload. The company also wants to identify the IP address a nd the timestamp details from the users so the company can prevent users from uploading pictures f rom unauthorized locations. Which solution will meet these requirements with LE AST development effort? A. Use AWS Panorama to identify celebrities in the p ictures. Use AWS CloudTrail to capture IP address and timestamp details.", + "options": [ + "B. Use AWS Panorama to identify celebrities in the p ictures. Make calls to the AWS Panorama Device", + "C. Use Amazon Rekognition to identify celebrities in the pictures. Use AWS CloudTrail to capture IP", + "D. Use Amazon Rekognition to identify celebrities in the pictures. Use the text detection feature to", + "A. Additionally, making calls to the AWS Panorama D evice" + ], + "correct": "C. Use Amazon Rekognition to identify celebrities in the pictures. Use AWS CloudTrail to capture IP", + "explanation": "The solution C will meet the requirements with the least development effort because it uses Amazon Rekognition and AWS CloudTrail, which are fully man aged services that can provide the desired functionality. The solution C involves the followin g steps: Use Amazon Rekognition to identify celebrities in t he pictures. Amazon Rekognition is a service that can analyze images and videos and extract insights such as faces, objects, scenes, emotions, and more. Amazon Rekognition also provides a feature ca lled Celebrity Recognition, which can recognize thousands of celebrities across a number of categor ies, such as politics, sports, entertainment, and medi a. Amazon Rekognition can return the name, face, an d confidence score of the recognized celebrities, as well as additional information such as URLs and biographies1. Use AWS CloudTrail to capture IP address and timest amp details. AWS CloudTrail is a service that can record the API calls and events made by or on behal f of AWS accounts. AWS CloudTrail can provide information such as the source IP address, the user identity, the request parameters, and the response elements of the API calls. AWS CloudTrail can also deliver the event records to an Amazon S3 bucket or an Amazon CloudWatch Logs group for fu rther analysis and auditing2. The other options are not suitable because: Option A: Using AWS Panorama to identify celebritie s in the pictures and using AWS CloudTrail to capture IP address and timestamp details will not m eet the requirements effectively. AWS Panorama is a service that can extend computer vision to the edge, where it can run inference on video streams from cameras and other devices. AWS Panorama is not designed for identifying celebrities in pictures, and it may not provide accurate or releva nt results. Moreover, AWS Panorama requires the use of an AWS Panorama Appliance or a compatible de vice, which may incur additional costs and complexity3. Option B: Using AWS Panorama to identify celebritie s in the pictures and making calls to the AWS Panorama Device SDK to capture IP address and times tamp details will not meet the requirements effectively, for the same reasons as option", + "references": "1: Amazon Rekognition Celebrity Recognition 2: AWS CloudTrail Overview 3: AWS Panorama Overview 4: AWS Panorama Device SDK 5: Amazon Rekognition Text Detection" + }, + { + "question": "using Amazon Kinesis Data Firehose. The company use s a small, server-based application in each store to send the data to AWS over the internet. Th e company uses this data to train a machine learning model that is retrained each day. The comp any's data science team has identified existing attributes on these records that could be combined to create an improved model. Which change will create the required transformed r ecords with the LEAST operational overhead?", + "options": [ + "A. Create an AWS Lambda function that can transform the incoming records. Enable data", + "B. Deploy an Amazon EMR cluster that runs Apache Spa rk and includes the transformation logic. Use", + "C. Deploy an Amazon S3 File Gateway in the stores. U pdate the in-store software to deliver data to", + "D. Launch a fleet of Amazon EC2 instances that inclu de the transformation logic. Configure the EC2" + ], + "correct": "A. Create an AWS Lambda function that can transform the incoming records. Enable data", + "explanation": "The solution A will create the required transformed records with the least operational overhead because it uses AWS Lambda and Amazon Kinesis Data Firehose, which are fully managed services that can provide the desired functionality. The sol ution A involves the following steps: Create an AWS Lambda function that can transform th e incoming records. AWS Lambda is a service that can run code without provisioning or managing servers. AWS Lambda can execute the transformation logic on the purchasing records and add the new attributes to the records1. Enable data transformation on the ingestion Kinesis Data Firehose delivery stream. Use the Lambda function as the invocation target. Amazon Kinesis D ata Firehose is a service that can capture, transform, and load streaming data into AWS data st ores. Amazon Kinesis Data Firehose can enable data transformation and invoke the Lambda function to process the incoming records before delivering them to Amazon S3. This can reduce the o perational overhead of managing the transformation process and the data storage2. The other options are not suitable because: Option B: Deploying an Amazon EMR cluster that runs Apache Spark and includes the transformation logic, using Amazon EventBridge (Amazon CloudWatch Events) to schedule an AWS Lambda function to launch the cluster each day and transform the re cords that accumulate in Amazon S3, and delivering the transformed records to Amazon S3 wil l incur more operational overhead than using AWS Lambda and Amazon Kinesis Data Firehose. The co mpany will have to manage the Amazon EMR cluster, the Apache Spark application, the AWS Lamb da function, and the Amazon EventBridge rule. Moreover, this solution will introduce a dela y in the transformation process, as it will run onl y once a day3. Option C: Deploying an Amazon S3 File Gateway in th e stores, updating the in-store software to deliver data to the S3 File Gateway, and using a sc heduled daily AWS Glue job to transform the data that the S3 File Gateway delivers to Amazon S3 will incur more operational overhead than using AWS Lambda and Amazon Kinesis Data Firehose. The compan y will have to manage the S3 File Gateway, the in-store software, and the AWS Glue job. Moreov er, this solution will introduce a delay in the transformation process, as it will run only once a day4. Option D: Launching a fleet of Amazon EC2 instances that include the transformation logic, configuring the EC2 instances with a daily cron job to transform the records that accumulate in Amazon S3, and delivering the transformed records t o Amazon S3 will incur more operational overhead than using AWS Lambda and Amazon Kinesis D ata Firehose. The company will have to manage the EC2 instances, the transformation code, and the cron job. Moreover, this solution will introduce a delay in the transformation process, as it will run only once a day5.", + "references": "1: AWS Lambda 2: Amazon Kinesis Data Firehose 3: Amazon EMR 4: Amazon S3 File Gateway 5: Amazon EC2" + }, + { + "question": "A company wants to segment a large group of custome rs into subgroups based on shared characteristics. The companys data scientist is pla nning to use the Amazon SageMaker built-in kmeans clustering algorithm for this task. The data scient ist needs to determine the optimal number of subgroups (k) to use. Which data visualization approach will MOST accurat ely determine the optimal value of k?", + "options": [ + "A. Calculate the principal component analysis (PCA) components. Run the k-means clustering", + "B. Calculate the principal component analysis (PCA) components. Create a line plot of the number of", + "C. Create a t-distributed stochastic neighbor embedd ing (t-SNE) plot for a range of perplexity values.", + "D. Run the k-means clustering algorithm for a range of k. For each value of k, calculate the sum of" + ], + "correct": "D. Run the k-means clustering algorithm for a range of k. For each value of k, calculate the sum of", + "explanation": "The solution D is the best data visualization appro ach to determine the optimal value of k for the kme ans clustering algorithm. The solution D involves the f ollowing steps: Run the k-means clustering algorithm for a range of k. For each value of k, calculate the sum of squared errors (SSE). The SSE is a measure of how w ell the clusters fit the data. It is calculated by summing the squared distances of each data point to its closest cluster center. A lower SSE indicates a better fit, but it will always decrease as the nu mber of clusters increases. Therefore, the goal is to find the smallest value of k that still has a low S SE1. Plot a line chart of the SSE for each value of k. T he line chart will show how the SSE changes as the value of k increases. Typically, the line chart wil l have a shape of an elbow, where the SSE drops rapidly at first and then levels off. The optimal v alue of k is the point after which the curve starts decreasing in a linear fashion. This point is also known as the elbow point, and it represents the balance between the number of clusters and the SSE1 . The other options are not suitable because: Option A: Calculating the principal component analy sis (PCA) components, running the k-means clustering algorithm for a range of k by using only the first two PCA components, and creating a scatter plot with a different color for each cluste r will not accurately determine the optimal value o f k. PCA is a technique that reduces the dimensionali ty of the data by transforming it into a new set of features that capture the most variance in the data . However, PCA may not preserve the original structure and distances of the data, and it may los e some information in the process. Therefore, running the k-means clustering algorithm on the PCA components may not reflect the true clusters in the data. Moreover, using only the first two PCA co mponents may not capture enough variance to represent the data well. Furthermore, creating a sc atter plot may not be reliable, as it depends on the subjective judgment of the data scientist to de cide when the clusters look reasonably separated2. Option B: Calculating the PCA components and creati ng a line plot of the number of components against the explained variance will not determine t he optimal value of k. This approach is used to determine the optimal number of PCA components to u se for dimensionality reduction, not for clustering. The explained variance is the ratio of the variance of each PCA component to the total variance of the data. The optimal number of PCA com ponents is the point where adding more components does not significantly increase the expl ained variance. However, this number may not correspond to the optimal number of clusters, as PC A and k-means clustering have different objectives and assumptions2. Option C: Creating a t-distributed stochastic neigh bor embedding (t-SNE) plot for a range of perplexity values will not determine the optimal va lue of k. t-SNE is a technique that reduces the dimensionality of the data by embedding it into a l ower-dimensional space, such as a twodimensional plane. t-SNE preserves the local structure and dist ances of the data, and it can reveal clusters and patterns in the data. However, t-SNE d oes not assign labels or centroids to the clusters, and it does not provide a measure of how well the c lusters fit the data. Therefore, t-SNE cannot determine the optimal number of clusters, as it onl y visualizes the data. Moreover, t-SNE depends on the perplexity parameter, which is a measure of how many neighbors each point considers. The perplexity parameter can affect the shape and size of the clusters, and there is no optimal value for it. Therefore, creating a t-SNE plot for a range of perplexity values may not be consistent or reliabl e3.", + "references": "1: How to Determine the Optimal K for K-Means? 2: Principal Component Analysis 3: t-Distributed Stochastic Neighbor Embedding" + }, + { + "question": "A car company is developing a machine learning solu tion to detect whether a car is present in an image. The image dataset consists of one million im ages. Each image in the dataset is 200 pixels in height by 200 pixels in width. Each image is labele d as either having a car or not having a car. Which architecture is MOST likely to produce a mode l that detects whether a car is present in an image with the highest accuracy?", + "options": [ + "A. Use a deep convolutional neural network (CNN) cla ssifier with the images as input. Include a", + "B. Use a deep convolutional neural network (CNN) cla ssifier with the images as input. Include a", + "C. Use a deep multilayer perceptron (MLP) classifier with the images as input. Include a linear output", + "D. Use a deep multilayer perceptron (MLP) classifier with the images as input. Include a softmax" + ], + "correct": "A. Use a deep convolutional neural network (CNN) cla ssifier with the images as input. Include a", + "explanation": "A deep convolutional neural network (CNN) classifie r is a suitable architecture for image classification tasks, as it can learn features from the images and reduce the dimensionality of the input. A linear output layer that outputs the proba bility that an image contains a car is appropriate for a binary classification problem, as it can prod uce a single scalar value between 0 and 1. A softma x output layer is more suitable for a multi-class cla ssification problem, as it can produce a vector of probabilities that sum up to 1. A deep multilayer p erceptron (MLP) classifier is not as effective as a CNN for image classification, as it does not exploi t the spatial structure of the images and requires a large number of parameters to process the high-dime nsional input. Reference: AWS Certified Machine Learning - Specialty Exam Gui de AWS Training - Machine Learning on AWS AWS Whitepaper - An Overview of Machine Learning on AWS", + "references": "" + }, + { + "question": "A data science team is working with a tabular datas et that the team stores in Amazon S3. The team wants to experiment with different feature transfor mations such as categorical feature encoding. Then the team wants to visualize the resulting dist ribution of the dataset. After the team finds an appropriate set of feature transformations, the tea m wants to automate the workflow for feature transformations. Which solution will meet these requirements with th e MOST operational efficiency?", + "options": [ + "A. Use Amazon SageMaker Data Wrangler preconfigured transformations to explore feature", + "B. Use an Amazon SageMaker notebook instance to expe riment with different feature", + "C. Use AWS Glue Studio with custom code to experimen t with different feature transformations. Save", + "D. Use Amazon SageMaker Data Wrangler preconfigured transformations to experiment with" + ], + "correct": "A. Use Amazon SageMaker Data Wrangler preconfigured transformations to explore feature", + "explanation": "The solution A will meet the requirements with the most operational efficiency because it uses Amazon SageMaker Data Wrangler, which is a service that simplifies the process of data preparation and feature engineering for machine learning. The s olution A involves the following steps: Use Amazon SageMaker Data Wrangler preconfigured tr ansformations to explore feature transformations. Amazon SageMaker Data Wrangler pro vides a visual interface that allows data scientists to apply various transformations to thei r tabular data, such as encoding categorical features, scaling numerical features, imputing miss ing values, and more. Amazon SageMaker Data Wrangler also supports custom transformations using Python code or SQL queries1. Use SageMaker Data Wrangler templates for visualiza tion. Amazon SageMaker Data Wrangler also provides a set of templates that can generate visua lizations of the data, such as histograms, scatter plots, box plots, and more. These visualizations ca n help data scientists to understand the distribution and characteristics of the data, and t o compare the effects of different feature transformations1. Export the feature processing workflow to a SageMak er pipeline for automation. Amazon SageMaker Data Wrangler can export the feature processing wor kflow as a SageMaker pipeline, which is a service that orchestrates and automates machine lea rning workflows. A SageMaker pipeline can run the feature processing steps as a preprocessing ste p, and then feed the output to a training step or an inference step. This can reduce the operational overhead of managing the feature processing workflow and ensure its consistency and reproducibi lity2. The other options are not suitable because: Option B: Using an Amazon SageMaker notebook instan ce to experiment with different feature transformations, saving the transformations to Amaz on S3, using Amazon QuickSight for visualization, and packaging the feature processing steps into an AWS Lambda function for automation will incur more operational overhead tha n using Amazon SageMaker Data Wrangler. The data scientist will have to write the code for the feature transformations, the data storage, the data visualization, and the Lambda function. Moreover, A WS Lambda has limitations on the execution time, memory size, and package size, which may not be sufficient for complex feature processing tasks3. Option C: Using AWS Glue Studio with custom code to experiment with different feature transformations, saving the transformations to Amaz on S3, using Amazon QuickSight for visualization, and packaging the feature processing steps into an AWS Lambda function for automation will incur more operational overhead tha n using Amazon SageMaker Data Wrangler. AWS Glue Studio is a visual interface that allows d ata engineers to create and run extract, transform,and load (ETL) jobs on AWS Glue. However, AWS Glue Studio does not provide preconfigured transformations or templates for feature engineerin g or data visualization. The data scientist will have to write custom code for these tasks, as well as for the Lambda function. Moreover, AWS Glue Studio is not integrated with SageMaker pipelines, and it may not be optimized for machine learning workflows4. Option D: Using Amazon SageMaker Data Wrangler prec onfigured transformations to experiment with different feature transformations, saving the transformations to Amazon S3, using Amazon QuickSight for visualization, packaging each featur e transformation step into a separate AWS Lambda function, and using AWS Step Functions for workflow automation will incur more operational overhead than using Amazon SageMaker Data Wrangler. The data scientist will have to create and manage multiple AWS Lambda functions and AWS Step F unctions, which can increase the complexity and cost of the solution. Moreover, AWS Lambda and AWS Step Functions may not be compatible with SageMaker pipelines, and they may not be optim ized for machine learning workflows5.", + "references": "1: Amazon SageMaker Data Wrangler 2: Amazon SageMaker Pipelines 3: AWS Lambda 4: AWS Glue Studio 5: AWS Step Functions" + }, + { + "question": "A company wants to conduct targeted marketing to se ll solar panels to homeowners. The company wants to use machine learning (ML) technologies to identify which houses already have solar panels. The company has collected 8,000 satellite images as training data and will use Amazon SageMaker Ground Truth to label the data. The company has a small internal team that is worki ng on the project. The internal team has no ML expertise and no ML experience. Which solution will meet these requirements with th e LEAST amount of effort from the internal team?", + "options": [ + "A. Set up a private workforce that consists of the i nternal team. Use the private workforce and the", + "B. Set up a private workforce that consists of the i nternal team. Use the private workforce to label", + "C. Set up a private workforce that consists of the i nternal team. Use the private workforce and the", + "D. Set up a public workforce. Use the public workfor ce to label the data. Use the SageMaker Object" + ], + "correct": "A. Set up a private workforce that consists of the i nternal team. Use the private workforce and the", + "explanation": "The solution A will meet the requirements with the least amount of effort from the internal team because it uses Amazon SageMaker Ground Truth and A mazon Rekognition Custom Labels, which are fully managed services that can provide the desired functionality. The solution A involves the following steps: Set up a private workforce that consists of the int ernal team. Use the private workforce and the SageMaker Ground Truth active learning feature to l abel the data. Amazon SageMaker Ground Truth is a service that can create high-quality training datasets for machine learning by using human labelers. A private workforce is a group of labeler s that the company can manage and control. The internal team can use the private workforce to labe l the satellite images as having solar panels or not. The SageMaker Ground Truth active learning fea ture can reduce the labeling effort by using a machine learning model to automatically label the e asy examples and only send the difficult ones to the human labelers1. Use Amazon Rekognition Custom Labels for model trai ning and hosting. Amazon Rekognition Custom Labels is a service that can train and deploy custo m machine learning models for image analysis. Amazon Rekognition Custom Labels can use the labele d data from SageMaker Ground Truth to train a model that can detect solar panels in satellite ima ges. Amazon Rekognition Custom Labels can also host the model and provide an API endpoint for infe rence2. The other options are not suitable because: Option B: Setting up a private workforce that consi sts of the internal team, using the private workforce to label the data, and using Amazon Rekog nition Custom Labels for model training and hosting will incur more effort from the internal te am than using SageMaker Ground Truth active learning feature. The internal team will have to la bel all the images manually, without the assistance of the machine learning model that can automate som e of the labeling tasks1. Option C: Setting up a private workforce that consi sts of the internal team, using the private workforce and the SageMaker Ground Truth active lea rning feature to label the data, using the SageMaker Object Detection algorithm to train a mod el, and using SageMaker batch transform for inference will incur more operational overhead than using Amazon Rekognition Custom Labels. The company will have to manage the SageMaker training job, the model artifact, and the batch transform job. Moreover, SageMaker batch transform is not suitable for real-time inference, as it processes the data in batches and stores the result s in Amazon S33. Option D: Setting up a public workforce, using the public workforce to label the data, using the SageMaker Object Detection algorithm to train a mod el, and using SageMaker batch transform for inference will incur more operational overhead and cost than using a private workforce and Amazon Rekognition Custom Labels. A public workforce is a group of labelers from Amazon Mechanical Turk, a crowdsourcing marketplace. The company will have to pay the public workforce for each labeling task, and it may not have full control over the qua lity and security of the labeled data. The company will also have to manage the SageMaker training job , the model artifact, and the batch transform job, as explained in option C4.", + "references": "1: Amazon SageMaker Ground Truth 2: Amazon Rekognition Custom Labels 3: Amazon SageMaker Object Detection 4: Amazon Mechanical Turk" + }, + { + "question": "A media company is building a computer vision model to analyze images that are on social media. The model consists of CNNs that the company trained by using images that the company stores in Amazon S3. The company used an Amazon SageMaker tra ining job in File mode with a single Amazon EC2 On-Demand Instance. Every day, the company updates the model by using a bout 10,000 images that the company has collected in the last 24 hours. The company configu res training with only one epoch. The company wants to speed up training and lower costs without the need to make any code changes. Which solution will meet these requirements?", + "options": [ + "A. Instead of File mode, configure the SageMaker tra ining job to use Pipe mode. Ingest the data from", + "B. Instead Of File mode, configure the SageMaker tra ining job to use FastFile mode with no Other", + "C. Instead Of On-Demand Instances, configure the Sag eMaker training job to use Spot Instances.", + "D. Instead Of On-Demand Instances, configure the Sag eMaker training job to use Spot Instances." + ], + "correct": "C. Instead Of On-Demand Instances, configure the Sag eMaker training job to use Spot Instances.", + "explanation": "The solution C will meet the requirements because i t uses Amazon SageMaker Spot Instances, which are unused EC2 instances that are available at up t o 90% discount compared to On-Demand prices. Amazon SageMaker Spot Instances can speed up traini ng and lower costs by taking advantage of the spare EC2 capacity. The company does not need to ma ke any code changes to use Spot Instances, as it can simply enable the managed spot training opti on in the SageMaker training job configuration. The company also does not need to im plement model checkpoints, as it is using only one epoch for training, which means the model will not resume from a previous state1. The other options are not suitable because: Option A: Configuring the SageMaker training job to use Pipe mode instead of File mode will not speed up training or lower costs significantly. Pip e mode is a data ingestion mode that streams data directly from S3 to the training algorithm, without copying the data to the local storage of the training instance. Pipe mode can reduce the startup time of the training job and the disk space usage, but it does not affect the computation time or the instance price. Moreover, Pipe mode may require some code changes to handle the streaming data, dep ending on the training algorithm2. Option B: Configuring the SageMaker training job to use FastFile mode instead of File mode will not speed up training or lower costs significantly. Fas tFile mode is a data ingestion mode that copies dat a from S3 to the local storage of the training instan ce in parallel with the training process. FastFile mode can reduce the startup time of the training jo b and the disk space usage, but it does not affect the computation time or the instance price. Moreove r, FastFile mode is only available for distributed training jobs that use multiple instances, which is not the case for the company3. Option D: Configuring the SageMaker training job to use Spot Instances and implementing model checkpoints will not meet the requirements without the need to make any code changes. Model checkpoints are a feature that allows the training job to save the model state periodically to S3, and resume from the latest checkpoint if the training j ob is interrupted. Model checkpoints can help to avoid losing the training progress and ensure the m odel convergence, but they require some code changes to implement the checkpointing logic and th e resuming logic4.", + "references": "1: Managed Spot Training - Amazon SageMaker 2: Pipe Mode - Amazon SageMaker 3: FastFile Mode - Amazon SageMaker 4: Checkpoints - Amazon SageMaker" + }, + { + "question": "A data scientist is working on a forecast problem b y using a dataset that consists of .csv files that are stored in Amazon S3. The files contain a timestamp variable in the following format: March 1st, 2020, 08:14pm - There is a hypothesis about seasonal differences in the dependent variable. This number could be higher or lower for weekdays because some days and hours present varying values, so the day of the week, month, or hour could be an important factor. As a result, the data scientist needs to transform the timestamp into weekdays, month, and day as thre e separate variables to conduct an analysis. Which solution requires the LEAST operational overh ead to create a new dataset with the added features?", + "options": [ + "A. Create an Amazon EMR cluster. Develop PySpark cod e that can read the timestamp variable as a", + "B. Create a processing job in Amazon SageMaker. Deve lop Python code that can read the timestamp", + "C. Create a new flow in Amazon SageMaker Data Wrangl er. Import the S3 file, use the Featurize", + "D. Create an AWS Glue job. Develop code that can rea d the timestamp variable as a string, transform" + ], + "correct": "C. Create a new flow in Amazon SageMaker Data Wrangl er. Import the S3 file, use the Featurize", + "explanation": "The solution C will create a new dataset with the a dded features with the least operational overhead because it uses Amazon SageMaker Data Wrangler, whi ch is a service that simplifies the process of data preparation and feature engineering for machin e learning. The solution C involves the following steps: Create a new flow in Amazon SageMaker Data Wrangler . A flow is a visual representation of the data preparation steps that can be applied to one or mor e datasets. The data scientist can create a new flow in the Amazon SageMaker Studio interface and i mport the S3 file as a data source1. Use the Featurize date/time transform to generate t he new variables. Amazon SageMaker Data Wrangler provides a set of preconfigured transforma tions that can be applied to the data with a few clicks. The Featurize date/time transform can parse a date/time column and generate new columns for the year, month, day, hour, minute, second, day of week, and day of year. The data scientist can use this transform to create the new variables from the timestamp variable2. Save the dataset as a new file in Amazon S3. Amazon SageMaker Data Wrangler can export the transformed dataset as a new file in Amazon S3, or as a feature store in Amazon SageMaker Feature Store. The data scientist can choose the output for mat and location of the new file3. The other options are not suitable because: Option A: Creating an Amazon EMR cluster and develo ping PySpark code that can read the timestamp variable as a string, transform and creat e the new variables, and save the dataset as a new file in Amazon S3 will incur more operational o verhead than using Amazon SageMaker Data Wrangler. The data scientist will have to manage th e Amazon EMR cluster, the PySpark application, and the data storage. Moreover, the data scientist will have to write custom code for the date/time parsing and feature generation, which may require m ore development effort and testing4. Option B: Creating a processing job in Amazon SageM aker and developing Python code that can read the timestamp variable as a string, transform and c reate the new variables, and save the dataset as a new file in Amazon S3 will incur more operational o verhead than using Amazon SageMaker Data Wrangler. The data scientist will have to manage th e processing job, the Python code, and the data storage. Moreover, the data scientist will have to write custom code for the date/time parsing and feature generation, which may require more developm ent effort and testing5. Option D: Creating an AWS Glue job and developing c ode that can read the timestamp variable as a string, transform and create the new variables, and save the dataset as a new file in Amazon S3 will incur more operational overhead than using Amazon S ageMaker Data Wrangler. The data scientist will have to manage the AWS Glue job, the code, and the data storage. Moreover, the data scientist will have to write custom code for the date/time pa rsing and feature generation, which may require more development effort and testing6.", + "references": "1: Amazon SageMaker Data Wrangler 2: Featurize Date/Time - Amazon SageMaker Data Wran gler 3: Exporting Data - Amazon SageMaker Data Wrangler 4: Amazon EMR 5: Processing Jobs - Amazon SageMaker 6: AWS Glue" + }, + { + "question": "An automotive company uses computer vision in its a utonomous cars. The company trained its object detection models successfully by using trans fer learning from a convolutional neural network (CNN). The company trained the models by using PyTo rch through the Amazon SageMaker SDK. The vehicles have limited hardware and compute powe r. The company wants to optimize the model to reduce memory, battery, and hardware consumption without a significant sacrifice in accuracy. Which solution will improve the computational effic iency of the models?", + "options": [ + "A. Use Amazon CloudWatch metrics to gain visibility into the SageMaker training weights, gradients,", + "B. Use Amazon SageMaker Ground Truth to build and ru n data labeling workflows. Collect a larger", + "C. Use Amazon SageMaker Debugger to gain visibility into the training weights, gradients, biases," + ], + "correct": "C. Use Amazon SageMaker Debugger to gain visibility into the training weights, gradients, biases,", + "explanation": "The solution C will improve the computational effic iency of the models because it uses Amazon SageMaker Debugger and pruning, which are technique s that can reduce the size and complexity of the convolutional neural network (CNN) models. The solution C involves the following steps: Use Amazon SageMaker Debugger to gain visibility in to the training weights, gradients, biases, and activation outputs. Amazon SageMaker Debugger is a service that can capture and analyze the tensors that are emitted during the training proces s of machine learning models. Amazon SageMaker Debugger can provide insights into the model perfor mance, quality, and convergence. Amazon SageMaker Debugger can also help to identify and di agnose issues such as overfitting, underfitting, vanishing gradients, and exploding gradients1. Compute the filter ranks based on the training info rmation. Filter ranking is a technique that can measure the importance of each filter in a convolut ional layer based on some criterion, such as the average percentage of zero activations or the L1-no rm of the filter weights. Filter ranking can help t o identify the filters that have little or no contrib ution to the model output, and thus can be removed without affecting the model accuracy2. Apply pruning to remove the low-ranking filters. Pr uning is a technique that can reduce the size and complexity of a neural network by removing the redu ndant or irrelevant parts of the network, such as neurons, connections, or filters. Pruning can he lp to improve the computational efficiency, memory usage, and inference speed of the model, as well as to prevent overfitting and improve generalization3. Set the new weights based on the pruned set of filt ers. After pruning, the model will have a smaller and simpler architecture, with fewer filters in eac h convolutional layer. The new weights of the model can be set based on the pruned set of filters , either by initializing them randomly or by finetu ning them from the original weights4. Run a new training job with the pruned model. The p runed model can be trained again with the same or a different dataset, using the same or a di fferent framework or algorithm. The new training job can use the same or a different configuration o f Amazon SageMaker, such as the instance type, the hyperparameters, or the data ingestion mode. Th e new training job can also use Amazon SageMaker Debugger to monitor and analyze the train ing process and the model quality5. The other options are not suitable because: Option A: Using Amazon CloudWatch metrics to gain v isibility into the SageMaker training weights, gradients, biases, and activation outputs will not be as effective as using Amazon SageMaker Debugger. Amazon CloudWatch is a service that can m onitor and observe the operational health and performance of AWS resources and applications. Amaz on CloudWatch can provide metrics, alarms, dashboards, and logs for various AWS services, incl uding Amazon SageMaker. However, Amazon CloudWatch does not provide the same level of granu larity and detail as Amazon SageMaker Debugger for the tensors that are emitted during th e training process of machine learning models. Amazon CloudWatch metrics are mainly focuse d on the resource utilization and the training progress, not on the model performance, quality, an d convergence6. Option B: Using Amazon SageMaker Ground Truth to bu ild and run data labeling workflows and collecting a larger labeled dataset with the labeli ng workflows will not improve the computational efficiency of the models. Amazon SageMaker Ground T ruth is a service that can create high-quality training datasets for machine learning by using hum an labelers. A larger labeled dataset can help to improve the model accuracy and generalization, but it will not reduce the memory, battery, and hardware consumption of the model. Moreover, a larg er labeled dataset may increase the training time and cost of the model7. Option D: Using Amazon SageMaker Model Monitor to g ain visibility into the ModelLatency metric and OverheadLatency metric of the model after the c ompany deploys the model and increasing the model learning rate will not improve the computatio nal efficiency of the models. Amazon SageMaker Model Monitor is a service that can monitor and ana lyze the quality and performance of machine learning models that are deployed on Amazon SageMak er endpoints. The ModelLatency metric and the OverheadLatency metric can measure the inferenc e latency of the model and the endpoint, respectively. However, these metrics do not provide any information about the training weights, gradients, biases, and activation outputs of the mo del, which are needed for pruning. Moreover, increasing the model learning rate will not reduce the size and complexity of the model, but it may affect the model convergence and accuracy.", + "references": "1: Amazon SageMaker Debugger 2: Pruning Convolutional Neural Networks for Resour ce Efficient Inference 3: Pruning Neural Networks: A Survey 4: Learning both Weights and Connections for Effici ent Neural Networks 5: Amazon SageMaker Training Jobs 6: Amazon CloudWatch Metrics for Amazon SageMaker 7: Amazon SageMaker Ground Truth : Amazon SageMaker Model Monitor" + }, + { + "question": "A chemical company has developed several machine le arning (ML) solutions to identify chemical process abnormalities. The time series values of in dependent variables and the labels are available for the past 2 years and are sufficient to accurate ly model the problem. The regular operation label is marked as 0. The abn ormal operation label is marked as 1 . Process abnormalities have a significant negative effect on the companys profits. The company must avoid these abnormalities. Which metrics will indicate an ML solution that wil l provide the GREATEST probability of detecting an abnormality?", + "options": [ + "A. Precision = 0.91", + "B. Precision = 0.61", + "C. Precision = 0.7", + "D. Precision = 0.98" + ], + "correct": "B. Precision = 0.61", + "explanation": "The metrics that will indicate an ML solution that will provide the greatest probability of detecting an abnormality are precision and recall. Precision is the ratio of true positives (TP) to the total numbe r of predicted positives (TP + FP), where FP is false positives. Recall is the ratio of true positives ( TP) to the total number of actual positives (TP + FN), whe re FN is false negatives. A high precision means that the ML solution has a low rate of false alarms , while a high recall means that the ML solution ha s a high rate of true detections. For the chemical co mpany, the goal is to avoid process abnormalities, which are marked as 1 in the labels. Therefore, the company needs an ML solution that has a high recall for the positive class, meaning that it can detect most of the abnormalities and minimize the false negatives. Among the four options, option B h as the highest recall for the positive class, which is 0.98. This means that the ML solution can detect 98% of the abnormalities and miss only 2%. Option B also has a reasonable precision for the po sitive class, which is 0.61. This means that the ML solution has a false alarm rate of 39%, which may b e acceptable for the company, depending on the cost and benefit analysis. The other options have l ower recall for the positive class, which means tha t they have higher false negative rates, which can be more detrimental for the company than false positive rates.", + "references": "1: AWS Certified Machine Learning - Specialty Exam Guide 2: AWS Training - Machine Learning on AWS 3: AWS Whitepaper - An Overview of Machine Learning on AWS 4: Precision and recall" + }, + { + "question": "A pharmaceutical company performs periodic audits o f clinical trial sites to quickly resolve critical findings. The company stores audit documents in tex t format. Auditors have requested help from a data science team to quickly analyze the documents. The auditors need to discover the 10 main topics within the documents to prioritize and distr ibute the review work among the auditing team members. Documents that describe adverse events mus t receive the highest priority. A data scientist will use statistical modeling to d iscover abstract topics and to provide a list of th e top words for each category to help the auditors assess the relevance of the topic. Which algorithms are best suited to this scenario? (Choose two.)", + "options": [ + "A. Latent Dirichlet allocation (LDA)", + "B. Random Forest classifier", + "C. Neural topic modeling (NTM)", + "D. Linear support vector machine" + ], + "correct": "", + "explanation": "The algorithms that are best suited to this scenari o are latent Dirichlet allocation (LDA) and neural topic modeling (NTM), as they are both unsupervised learning methods that can discover abstract topics from a collection of text documents. LDA and NTM can provide a list of the top words for each topic, as well as the topic distribution for each d ocument, which can help the auditors assess the relevance and priority of the topic12. The other options are not suitable because: Option B: A random forest classifier is a supervise d learning method that can perform classification or regression tasks by using an ensemble of decisio n trees. A random forest classifier is not suitable for discovering abstract topics from text documents , as it requires labeled data and predefined classes3. Option D: A linear support vector machine is a supe rvised learning method that can perform classification or regression tasks by using a linea r function that separates the data into different classes. A linear support vector machine is not sui table for discovering abstract topics from text documents, as it requires labeled data and predefin ed classes4. Option E: A linear regression is a supervised learn ing method that can perform regression tasks by using a linear function that models the relationshi p between a dependent variable and one or more independent variables. A linear regression is not s uitable for discovering abstract topics from text documents, as it requires labeled data and a contin uous output variable5.", + "references": "1: Latent Dirichlet Allocation 2: Neural Topic Modeling 3: Random Forest Classifier 4: Linear Support Vector Machine 5: Linear Regression" + }, + { + "question": "A company wants to predict the classification of do cuments that are created from an application. New documents are saved to an Amazon S3 bucket ever y 3 seconds. The company has developed three versions of a machine learning (ML) model wit hin Amazon SageMaker to classify document text. The company wants to deploy these three versi ons to predict the classification of each document. Which approach will meet these requirements with th e LEAST operational overhead?", + "options": [ + "A. Configure an S3 event notification that invokes a n AWS Lambda function when new documents are created. Configure the Lambda function to creat e three SageMaker batch transform jobs, one", + "B. Deploy all the models to a single SageMaker endpo int. Treat each model as a production variant.", + "C. Deploy each model to its own SageMaker endpoint C onfigure an S3 event notification that invokes", + "D. Deploy each model to its own SageMaker endpoint. Create three AWS Lambda functions." + ], + "correct": "B. Deploy all the models to a single SageMaker endpo int. Treat each model as a production variant.", + "explanation": "The approach that will meet the requirements with t he least operational overhead is to deploy all the models to a single SageMaker endpoint, treat ea ch model as a production variant, configure an S3 event notification that invokes an AWS Lambda fu nction when new documents are created, and configure the Lambda function to call each producti on variant and return the results of each model. This approach involves the following steps: Deploy all the models to a single SageMaker endpoin t. Amazon SageMaker is a service that can build, train, and deploy machine learning models. A mazon SageMaker can deploy multiple models to a single endpoint, which is a web service that can serve predictions from the models. Each model can be treated as a production variant, which is a vers ion of the model that runs on one or more instances. Amazon SageMaker can distribute the traf fic among the production variants according to the specified weights1. Treat each model as a production variant. Amazon Sa geMaker can deploy multiple models to a single endpoint, which is a web service that can serve pre dictions from the models. Each model can be treated as a production variant, which is a version of the model that runs on one or more instances. Amazon SageMaker can distribute the traf fic among the production variants according to the specified weights1. Configure an S3 event notification that invokes an AWS Lambda function when new documents are created. Amazon S3 is a service that can store and retrieve any amount of data. Amazon S3 can send event notifications when certain actions occur on t he objects in a bucket, such as object creation, deletion, or modification. Amazon S3 can invoke an AWS Lambda function as a destination for the event notifications. AWS Lambda is a service that c an run code without provisioning or managing servers2. Configure the Lambda function to call each producti on variant and return the results of each model. AWS Lambda can execute the code that can call the S ageMaker endpoint and specify the production variant to invoke. AWS Lambda can use the AWS SDK o r the SageMaker Runtime API to send requests to the endpoint and receive the predictions from th e models. AWS Lambda can return the results of each model as a response to the event notification3 . The other options are not suitable because: Option A: Configuring an S3 event notification that invokes an AWS Lambda function when new documents are created, configuring the Lambda funct ion to create three SageMaker batch transform jobs, one batch transform job for each model for ea ch document, will incur more operational overhead than using a single SageMaker endpoint. Am azon SageMaker batch transform is a service that can process large datasets in batches and stor e the predictions in Amazon S3. Amazon SageMaker batch transform is not suitable for real- time inference, as it introduces a delay between the request and the response. Moreover, creating th ree batch transform jobs for each document will increase the complexity and cost of the solution4. Option C: Deploying each model to its own SageMaker endpoint, configuring an S3 event notification that invokes an AWS Lambda function when new docume nts are created, configuring the Lambda function to call each endpoint and return the resul ts of each model, will incur more operational overhead than using a single SageMaker endpoint. De ploying each model to its own endpoint will increase the number of resources and endpoints to m anage and monitor. Moreover, calling each endpoint separately will increase the latency and n etwork traffic of the solution5. Option D: Deploying each model to its own SageMaker endpoint, creating three AWS Lambda functions, configuring each Lambda function to call a different endpoint and return the results, configuring three S3 event notifications to invoke the Lambda functions when new documents are created, will incur more operational overhead than using a single SageMaker endpoint and a single Lambda function. Deploying each model to its own en dpoint will increase the number of resources and endpoints to manage and monitor. Creating three Lambda functions will increase the complexity and cost of the solution. Configuring three S3 even t notifications will increase the number of trigger s and destinations to manage and monitor6.", + "references": "1: Deploying Multiple Models to a Single Endpoint - Amazon SageMaker 2: Configuring Amazon S3 Event Notifications - Amaz on Simple Storage Service 3: Invoke an Endpoint - Amazon SageMaker 4: Get Inferences for an Entire Dataset with Batch Transform - Amazon SageMaker 5: Deploy a Model - Amazon SageMaker 6: AWS Lambda" + }, + { + "question": "A company wants to detect credit card fraud. The co mpany has observed that an average of 2% of credit card transactions are fraudulent. A data sci entist trains a classifier on a year's worth of cre dit card transaction dat", + "options": [ + "A. The classifier needs to identify the fraudulent t ransactions. The company wants to accurately", + "B. Specificity", + "C. False positive rate", + "D. Accuracy" + ], + "correct": "", + "explanation": "The F1 score is a measure of the harmonic mean of p recision and recall, which are both important for fraud detection. Precision is the ratio of true positives to all predicted positives, and recall i s the ratio of true positives to all actual positives. A high F1 score indicates that the classifier can cor rectly identify fraudulent transactions and avoid false ne gatives. The true positive rate is another name for recall, and it measures the proportion of fraudulen t transactions that are correctly detected by the classifier. A high true positive rate means that th e classifier can capture as many fraudulent transactions as possible.", + "references": "Fraud Detection Using Machine Learning | Implementa tions | AWS Solutions Detect fraudulent transactions using machine learni ng with Amazon SageMaker | AWS Machine Learning Blog 1. Introduction \" Reproducible Machine Learning for Credit Card Fraud Detection" + }, + { + "question": "Each morning, a data scientist at a rental car comp any creates insights about the previous days rental car reservation demands. The company needs t o automate this process by streaming the data to Amazon S3 in near real time. The solution must d etect high-demand rental cars at each of the companys locations. The solution also must create a visualization dashboard that automatically refreshes with the most recent data. Which solution will meet these requirements with th e LEAST development time? A. Use Amazon Kinesis Data Firehose to stream the re servation data directly to Amazon S3. Detect high-demand outliers by using Amazon QuickSight ML Insights. Visualize the data in QuickSight.", + "options": [ + "B. Use Amazon Kinesis Data Streams to stream the res ervation data directly to Amazon S3. Detect", + "C. Use Amazon Kinesis Data Firehose to stream the re servation data directly to Amazon S3. Detect", + "D. Use Amazon Kinesis Data Streams to stream the res ervation data directly to Amazon S3. Detect" + ], + "correct": "", + "explanation": "The solution that will meet the requirements with t he least development time is to use Amazon Kinesis Data Firehose to stream the reservation dat a directly to Amazon S3, detect high-demand outliers by using Amazon QuickSight ML Insights, an d visualize the data in QuickSight. This solution does not require any custom development or ML domai n expertise, as it leverages the built-in features of QuickSight ML Insights to automatically run anomaly detection and generate insights on the streaming data. QuickSight ML Insights can also create a visualization dashboard that automatically refreshes with the most recent data, and allows the data scientist to explore the outliers and their key drivers. Reference: 1: Simplify and automate anomaly detection in strea ming data with Amazon Lookout for Metrics | AWS Machine Learning Blog 2: Detecting outliers with ML-powered anomaly detec tion - Amazon QuickSight 3: Real-time Outlier Detection Over Streaming Data - IEEE Xplore 4: Towards a deep learning-based outlier detection \u00a6 - Journal of Big Data", + "references": "" + }, + { + "question": "A network security vendor needs to ingest telemetry data from thousands of endpoints that run all over the world. The data is transmitted every 30 se conds in the form of records that contain 50 fields . Each record is up to 1 KB in size. The security ven dor uses Amazon Kinesis Data Streams to ingest the dat", + "options": [ + "A. The vendor requires hourly summaries of the recor ds that Kinesis Data Streams ingests. The", + "B. Use AWS Lambda to read and aggregate the data hou rly. Transform the data and store it in", + "C. Use Amazon Kinesis Data Firehose to read and aggr egate the data hourly. Transform the data and", + "D. Use Amazon Kinesis Data Analytics to read and agg regate the data hourly. Transform the data and" + ], + "correct": "C. Use Amazon Kinesis Data Firehose to read and aggr egate the data hourly. Transform the data and", + "explanation": "Explanation/Reference: The solution that will meet the requirements with t he least amount of customization to transform and store the ingested data is to use Amazon Kinesi s Data Analytics to read and aggregate the data hourly, transform the data and store it in Amazon S 3 by using Amazon Kinesis Data Firehose. This solution leverages the built-in features of Kinesis Data Analytics to perform SQL queries on streaming data and generate hourly summaries. Kinesis Data An alytics can also output the transformed data to Kinesis Data Firehose, which can then deliver the d ata to S3 in a specified format and partitioning scheme. This solution does not require any custom c ode or additional infrastructure to process the data. The other solutions either require more custo mization (such as using Lambda or EMR) or do not meet the requirement of aggregating the data hourly (such as using Lambda to read the data from Kinesis Data Streams). Reference: 1: Boosting Resiliency with an ML-based Telemetry A nalytics Architecture | AWS Architecture Blog 2: AWS Cloud Data Ingestion Patterns and Practices 3: IoT ingestion and Machine Learning analytics pip eline with AWS IoT \u00a6 4: AWS IoT Data Ingestion Simplified 101: The Compl ete Guide - Hevo Data", + "references": "" + }, + { + "question": "A machine learning (ML) specialist uploads 5 TB of data to an Amazon SageMaker Studio environment. The ML specialist performs initial dat a cleansing. Before the ML specialist begins to train a model, the ML specialist needs to create an d view an analysis report that details potential bi as in the uploaded data. Which combination of actions will meet these requir ements with the LEAST operational overhead? (Choose two.)", + "options": [ + "A. Use SageMaker Clarify to automatically detect dat a bias", + "B. Turn on the bias detection option in SageMaker Gr ound Truth to automatically analyze data", + "C. Use SageMaker Model Monitor to generate a bias dr ift report.", + "D. Configure SageMaker Data Wrangler to generate a b ias report." + ], + "correct": "", + "explanation": "The combination of actions that will meet the requi rements with the least operational overhead is to use SageMaker Clarify to automatically detect data bias and to configure SageMaker Data Wrangler to generate a bias report. SageMaker Clarify is a f eature of Amazon SageMaker that provides machine learning (ML) developers with tools to gain greater insights into their ML training data and models. SageMaker Clarify can detect potential bias during data preparation, after model training, and in your deployed model. For instance, you can c heck for bias related to age in your dataset or in your trained model and receive a detailed report th at quantifies different types of potential bias1. SageMaker Data Wrangler is another feature of Amazo n SageMaker that enables you to prepare data for machine learning (ML) quickly and easily. You c an use SageMaker Data Wrangler to identify potential bias during data preparation without havi ng to write your own code. You specify input features, such as gender or age, and SageMaker Data Wrangler runs an analysis job to detect potential bias in those features. SageMaker Data Wr angler then provides a visual report with a description of the metrics and measurements of pote ntial bias so that you can identify steps to remediate the bias2. The other actions either requi re more customization (such as using SageMaker Model Monitor or SageMaker Experiments) or do not m eet the requirement of detecting data bias (such as using SageMaker Ground Truth). Reference: 1: Bias Detection and Model Explainability \" Amazon Web Services 2: Amazon SageMaker Data Wrangler \" Amazon Web Serv ices", + "references": "" + }, + { + "question": "A medical device company is building a machine lear ning (ML) model to predict the likelihood of device recall based on customer data that the compa ny collects from a plain text survey. One of the survey questions asks which medications the custome r is taking. The data for this field contains the names of medications that customers enter manually. Customers misspell some of the medication names. The column that contains the medication name data gives a categorical feature with high cardinality but redundancy. What is the MOST effective way to encode this categ orical feature into a numeric feature?", + "options": [ + "A. Spell check the column. Use Amazon SageMaker one- hot encoding on the column to transform a", + "B. Fix the spelling in the column by using char-RNN. Use Amazon SageMaker Data Wrangler one-hot", + "C. Use Amazon SageMaker Data Wrangler similarity enc oding on the column to create embeddings", + "D. Use Amazon SageMaker Data Wrangler ordinal encodi ng on the column to encode categories into" + ], + "correct": "C. Use Amazon SageMaker Data Wrangler similarity enc oding on the column to create embeddings", + "explanation": "The most effective way to encode this categorical f eature into a numeric feature is to use Amazon SageMaker Data Wrangler similarity encoding on the column to create embeddings of vectors of real numbers. Similarity encoding is a technique that tr ansforms categorical features into numerical features by computing the similarity between the ca tegories. Similarity encoding can handle high cardinality and redundancy in categorical features, as it can group similar categories together based on their string similarity. For example, if the col umn contains the values oeaspirin , oeasprin , and oeibuprofen , similarity encoding will assign a hig h similarity score to oeaspirin and oeasprin , and a low similarity score to oeibuprofen . Similarity encodi ng can also create embeddings of vectors of real numbers, which can be used as input for machine lea rning models. Amazon SageMaker Data Wrangler is a feature of Amazon SageMaker that enab les you to prepare data for machine learning quickly and easily. You can use SageMaker Data Wran gler to apply similarity encoding to a column of categorical data, and generate embeddings of vector s of real numbers that capture the similarity between the categories1. The other options are eith er less effective or more complex to implement. Spell checking the column and using one-hot encodin g would require additional steps and resources, and may not capture all the misspellings or redunda ncies. One-hot encoding would also create a large number of features, which could increase the dimensionality and sparsity of the dat a. Ordinal encoding would assign an arbitrary order to the categories, which could introduce bias or noise in the data. Reference: 1: Amazon SageMaker Data Wrangler \" Amazon Web Serv ices", + "references": "" + }, + { + "question": "A manufacturing company wants to create a machine l earning (ML) model to predict when equipment is likely to fail. A data science team al ready constructed a deep learning model by using TensorFlow and a custom Python script in a local en vironment. The company wants to use Amazon SageMaker to train the model. Which TensorFlow estimator configuration will train the model MOST cost-effectively?", + "options": [ + "A. Turn on SageMaker Training Compiler by adding com piler_config=TrainingCompilerConfig() as a", + "B. Turn on SageMaker Training Compiler by adding com piler_config=TrainingCompilerConfig() as a", + "C. Adjust the training script to use distributed dat a parallelism. Specify appropriate values for the", + "D. Turn on SageMaker Training Compiler by adding com piler_config=TrainingCompilerConfig() as a" + ], + "correct": "", + "explanation": "The TensorFlow estimator configuration that will tr ain the model most cost-effectively is to turn on SageMaker Training Compiler by adding compiler_conf ig=TrainingCompilerConfig() as a parameter, turn on managed spot training by setting the use_sp ot_instances parameter to True, and pass the script to the estimator in the call to the TensorFl ow fit() method. This configuration will optimize t he model for the target hardware platform, reduce the training cost by using Amazon EC2 Spot Instances, and use the custom Python script without any modification. SageMaker Training Compiler is a feature of Amazon SageMaker that enables you to optimize your TensorFlow, PyTorch, and MXNet models for inference on a variety of target hardware platforms. SageMaker Training Compiler can improve the inferen ce performance and reduce the inference cost of your models by applying various compilation tech niques, such as operator fusion, quantization, pruning, and graph optimization. You can enable Sag eMaker Training Compiler by adding compiler_config=TrainingCompilerConfig() as a param eter to the TensorFlow estimator constructor1. Managed spot training is another feature of Amazon SageMaker that enables you to use Amazon EC2 Spot Instances for training your machine learning m odels. Amazon EC2 Spot Instances let you take advantage of unused EC2 capacity in the AWS Cloud. Spot Instances are available at up to a 90% discount compared to On-Demand prices. You can use Spot Instances for various fault-tolerant and flexible applications. You can enable managed spot training by setting the use_spot_instances parameter to True and specifying the max_wait and m ax_run parameters in the TensorFlow estimator constructor2. The TensorFlow estimator is a class in the SageMake r Python SDK that allows you to train and deploy TensorFlow models on SageMaker. You can use the Ten sorFlow estimator to run your own Python script on SageMaker, without any modification. You can pass the script to the estimator in the call to the TensorFlow fit() method, along with the locatio n of your input data. The fit() method starts a SageMaker training job and runs your script as the entry point in the training containers3. The other options are either less cost-effective or more complex to implement. Adjusting the training script to use distributed data parallelism would re quire modifying the script and specifying appropriate values for the distribution parameter, which could increase the development time and complexity. Setting the MaxWaitTimeInSeconds parame ter to be equal to the MaxRuntimeInSeconds parameter would not reduce the cost, as it would on ly specify the maximum duration of the training job, regardless of the instance type.", + "references": "1: Optimize TensorFlow, PyTorch, and MXNet models f or deployment using Amazon SageMaker Training Compiler | AWS Machine Learning Blog 2: Managed Spot Training: Save Up to 90% On Your Am azon SageMaker Training Jobs | AWS Machine Learning Blog 3: sagemaker.tensorflow \" sagemaker 2.66.0 document ation" + }, + { + "question": "A company is creating an application to identify, c ount, and classify animal images that are uploaded to the companys website. The company is using the A mazon SageMaker image classification algorithm with an ImageNetV2 convolutional neural n etwork (CNN). The solution works well for most animal images but does not recognize many anim al species that are less common. The company obtains 10,000 labeled images of less c ommon animal species and stores the images in Amazon S3. A machine learning (ML) engineer needs t o incorporate the images into the model by using Pipe mode in SageMaker. Which combination of steps should the ML engineer t ake to train the model? (Choose two.)", + "options": [ + "A. Use a ResNet model. Initiate full training mode b y initializing the network with random weights.", + "B. Use an Inception model that is available with the SageMaker image classification algorithm.", + "C. Create a .lst file that contains a list of image files and corresponding class labels. Upload the .l st file", + "D. Initiate transfer learning. Train the model by usin g the images of less common species. E. Use an augmented manifest file in JSON Lines format ." + ], + "correct": "", + "explanation": "The combination of steps that the ML engineer shoul d take to train the model are to create a .lst file that contains a list of image files and correspondi ng class labels, upload the .lst file to Amazon S3, and initiate transfer learning by training the mode l using the images of less common species. This approach will allow the ML engineer to leverage the existing ImageNetV2 CNN model and fine-tune it with the new data using Pipe mode in SageMaker. A .lst file is a text file that contains a list of image files and corresponding class labels, separat ed by tabs. The .lst file format is required for using th e SageMaker image classification algorithm with Pip e mode. Pipe mode is a feature of SageMaker that enab les streaming data directly from Amazon S3 to the training instances, without downloading the dat a first. Pipe mode can reduce the startup time, improve the I/O throughput, and enable training on large datasets that exceed the disk size limit. To use Pipe mode, the ML engineer needs to upload the .lst file to Amazon S3 and specify the S3 path as the input data channel for the training job1. Transfer learning is a technique that enables reusi ng a pre-trained model for a new task by finetuning the model parameters with new data. Transfer learni ng can save time and computational resources, as well as improve the performance of th e model, especially when the new task is similar to the original task. The SageMaker image classific ation algorithm supports transfer learning by allowing the ML engineer to specify the number of o utput classes and the number of layers to be retrained. The ML engineer can use the existing Ima geNetV2 CNN model, which is trained on 1,000 classes of common objects, and fine-tune it with th e new data of less common animal species, which is a similar task2. The other options are either less effective or not supported by the SageMaker image classification algorithm. Using a ResNet model and initiating full training mode would require training the model from scratch, which would take more time and resour ces than transfer learning. Using an Inception model is not possible, as the SageMaker image class ification algorithm only supports ResNet and ImageNetV2 models. Using an augmented manifest file in JSON Lines format is not compatible with Pipe mode, as Pipe mode only supports .lst files fo r image classification1.", + "references": "1: Using Pipe input mode for Amazon SageMaker algor ithms | AWS Machine Learning Blog 2: Image Classification Algorithm - Amazon SageMake r" + }, + { + "question": "A credit card company wants to identify fraudulent transactions in real time. A data scientist builds a machine learning model for this purpose. The transa ctional data is captured and stored in Amazon S3. The historic data is already labeled with two c lasses: fraud (positive) and fair transactions (negative). The data scientist removes all the miss ing data and builds a classifier by using the XGBoost algorithm in Amazon SageMaker. The model pr oduces the following results: True positive rate (TPR): 0.700 False negative rate (FNR): 0.300 True negative rate (TNR): 0.977 False positive rate (FPR): 0.023 Overall accuracy: 0.949 Which solution should the data scientist use to imp rove the performance of the model?", + "options": [ + "A. Apply the Synthetic Minority Oversampling Techniq ue (SMOTE) on the minority class in the", + "B. Apply the Synthetic Minority Oversampling Techniq ue (SMOTE) on the majority class in the", + "C. Undersample the minority class.", + "D. Oversample the majority class." + ], + "correct": "", + "explanation": "The solution that the data scientist should use to improve the performance of the model is to apply the Synthetic Minority Oversampling Technique (SMOT E) on the minority class in the training dataset, and retrain the model with the updated tra ining data. This solution can address the problem of class imbalance in the dataset, which can affect the models ability to learn from the rare but important positive class (fraud). Class imbalance is a common issue in machine learni ng, especially for classification tasks. It occurs when one class (usually the positive or target clas s) is significantly underrepresented in the dataset compared to the other class (usually the negative o r non-target class). For example, in the credit car d fraud detection problem, the positive class (fraud) is much less frequent than the negative class (fai r transactions). This can cause the model to be biase d towards the majority class, and fail to capture the characteristics and patterns of the minority cl ass. As a result, the model may have a high overall accuracy, but a low recall or true positive rate fo r the minority class, which means it misses many fraudulent transactions. SMOTE is a technique that can help mitigate the cla ss imbalance problem by generating synthetic samples for the minority class. SMOTE works by find ing the k-nearest neighbors of each minority class instance, and randomly creating new instances along the line segments connecting them. This way, SMOTE can increase the number and diversity of the minority class instances, without duplicating or losing any information. By applying SMOTE on the minority class in the training dataset, the data scientist can balance the classes and improve the models performance on the positive class1. The other options are either ineffective or counter productive. Applying SMOTE on the majority class would not balance the classes, but increase the imb alance and the size of the dataset. Undersampling the minority class would reduce the n umber of instances available for the model to learn from, and potentially lose some important inf ormation. Oversampling the majority class would also increase the imbalance and the size of the dat aset, and introduce redundancy and overfitting.", + "references": "1: SMOTE for Imbalanced Classification with Python - Machine Learning Mastery" + }, + { + "question": "A company processes millions of orders every day. T he company uses Amazon DynamoDB tables to store order information. When customers submit new orders, the new orders are immediately added to the DynamoDB tables. New orders arrive in the Dy namoDB tables continuously. A data scientist must build a peak-time prediction solution. The data scientist must also create an Amazon OuickSight dashboard to display near real-li me order insights. The data scientist needs to build a solution that will give QuickSight access t o the data as soon as new order information arrives . Which solution will meet these requirements with th e LEAST delay between when a new order is processed and when QuickSight can access the new or der information?", + "options": [ + "A. Use AWS Glue to export the data from Amazon Dynam oDB to Amazon S3. Configure OuickSight to", + "B. Use Amazon Kinesis Data Streams to export the dat a from Amazon DynamoDB to Amazon S3.", + "C. Use an API call from OuickSight to access the dat a that is in Amazon DynamoDB directly", + "D. Use Amazon Kinesis Data Firehose to export the da ta from Amazon DynamoDB to Amazon S3." + ], + "correct": "B. Use Amazon Kinesis Data Streams to export the dat a from Amazon DynamoDB to Amazon S3.", + "explanation": "The best solution for this scenario is to use Amazo n Kinesis Data Streams to export the data from Amazon DynamoDB to Amazon S3, and then configure Qu ickSight to access the data in Amazon S3. This solution has the following advantages: It allows near real-time data ingestion from Dynamo DB to S3 using Kinesis Data Streams, which can capture and process data continuously and at scale1 . It enables QuickSight to access the data in S3 usin g the Athena connector, which supports federated queries to multiple data sources, including Kinesis Data Streams2. It avoids the need to create and manage a Lambda fu nction or a Glue crawler, which are required for the other solutions. The other solutions have the following drawbacks: Using AWS Glue to export the data from DynamoDB to S3 introduces additional latency and complexity, as Glue is a batch-oriented service tha t requires scheduling and configuration3. Using an API call from QuickSight to access the dat a in DynamoDB directly is not possible, as QuickSight does not support direct querying of Dyna moDB4. Using Kinesis Data Firehose to export the data from DynamoDB to S3 is less efficient and flexible than using Kinesis Data Streams, as Firehose does not su pport custom data processing or transformation, and has a minimum buffer interval of 60 seconds5.", + "references": "1: Amazon Kinesis Data Streams - Amazon Web Service s 2: Visualize Amazon DynamoDB insights in Amazon Qui ckSight using the Amazon Athena DynamoDB connector and AWS Glue | AWS Big Data Blog 3: AWS Glue - Amazon Web Services 4: Visualising your Amazon DynamoDB data with Amazo n QuickSight - DEV Community 5: Amazon Kinesis Data Firehose - Amazon Web Servic es" + }, + { + "question": "A retail company wants to build a recommendation sy stem for the company's website. The system needs to provide recommendations for existing users and needs to base those recommendations on each user's past browsing history. The system also must filter out any items that the user previously purchased. Which solution will meet these requirements with th e LEAST development effort?", + "options": [ + "A. Train a model by using a user-based collaborative filtering algorithm on Amazon SageMaker. Host", + "B. Use an Amazon Personalize PERSONALIZED_RANKING re cipe to train a model. Create a real-time", + "C. Use an Amazon Personalize USER_ PERSONAL IZATION recipe to train a model Create a real-time", + "D. Train a neural collaborative filtering model on A mazon SageMaker by using GPU instances. Host" + ], + "correct": "C. Use an Amazon Personalize USER_ PERSONAL IZATION recipe to train a model Create a real-time", + "explanation": "Amazon Personalize is a fully managed machine learn ing service that makes it easy for developers to create personalized user experiences at scale. It u ses the same recommender system technology that Amazon uses to create its own personalized recommen dations. Amazon Personalize provides several pre-built recipes that can be used to train models for different use cases. The USER_PERSONALIZATION recipe is designed to provide personalized recommendations for existing users based on their past interactions with items. The PERSONALIZED_RANKING recipe is designed to re-rank a list of items for a user based on their p references. The USER_PERSONALIZATION recipe is more suitable for this use case because it can gene rate recommendations for each user without requiring a list of candidate items. To filter out the items that the user previously purchased, a rea ltime filter can be created and applied to the campaign. A real-time filter is a dynamic filter that uses the latest interaction data to exclude items from t he recommendations. By using Amazon Personalize, the development effort is minimized be cause it handles the data processing, model training, and deployment automatically. The web app lication can use the GetRecommendations API operation to get the real-time recommendations from the campaign. Reference: Amazon Personalize What is Amazon Personalize? USER_PERSONALIZATION recipe PERSONALIZED_RANKING recipe Filtering recommendations GetRecommendations API operation", + "references": "" + }, + { + "question": "A data engineer is preparing a dataset that a retai l company will use to predict the number of visitor s to stores. The data engineer created an Amazon S3 b ucket. The engineer subscribed the S3 bucket to an AWS Data Exchange data product for general econo mic indicators. The data engineer wants to join the economic indicator data to an existing table in Amazon Athena to merge with the business dat", + "options": [ + "A. All these transformations must finish running in 30-60 minutes.", + "B. Configure the AWS Data Exchange product as a prod ucer for an Amazon Kinesis data stream. Use", + "C. Use an S3 event on the AWS Data Exchange S3 bucke t to invoke an AWS Lambda function. Program", + "D. Use an S3 event on the AWS Data Exchange S3 bucke t to invoke an AWS Lambda Function Program" + ], + "correct": "B. Configure the AWS Data Exchange product as a prod ucer for an Amazon Kinesis data stream. Use", + "explanation": "The most cost-effective solution is to use an S3 ev ent to trigger a Lambda function that uses SageMaker Data Wrangler to merge the data. This sol ution avoids the need to provision and manage any additional resources, such as Kinesis streams, Firehose delivery streams, Glue jobs, or Redshift clusters. SageMaker Data Wrangler provides a visual interface to import, prepare, transform, and analyze data from various sources, including AWS Da ta Exchange products. It can also export the data preparation workflow to a Python script that can be executed by a Lambda function. This solution can meet the time requirement of 30-60 minutes, dependi ng on the size and complexity of the data.", + "references": "Using Amazon S3 Event Notifications Prepare ML Data with Amazon SageMaker Data Wrangler AWS Lambda Function" + }, + { + "question": "A social media company wants to develop a machine l earning (ML) model to detect Inappropriate or offensive content in images. The company has collec ted a large dataset of labeled images and plans to use the built-in Amazon SageMaker image classifi cation algorithm to train the model. The company also intends to use SageMaker pipe mode to speed up the training. ...company splits the dataset into training, valida tion, and testing datasets. The company stores the training and validation images in folders that are named Training and Validation, respectively. The folder ...ain subfolders that correspond to the nam es of the dataset classes. The company resizes the images to the same sue and generates two input mani fest files named training.1st and validation.1st, for the ..ing dataset and the validation dataset. r espectively. Finally, the company creates two separate Amazon S3 buckets for uploads of the train ing dataset and the validation dataset. ...h additional data preparation steps should the c ompany take before uploading the files to Amazon S3?", + "options": [ + "A. Generate two Apache Parquet files, training.parqu et and validation.parquet. by reading the", + "B. Compress the training and validation directories by using the Snappy compression library Upload", + "C. Compress the training and validation directories by using the gzip compression library. Upload the", + "D. Generate two RecordIO files, training rec and val idation.rec. from the manifest files by using the" + ], + "correct": "D. Generate two RecordIO files, training rec and val idation.rec. from the manifest files by using the", + "explanation": "The SageMaker image classification algorithm suppor ts both RecordIO and image content types for training in file mode, and supports the RecordIO co ntent type for training in pipe mode1. However, the algorithm also supports training in pipe mode u sing the image files without creating RecordIO files, by using the augmented manifest format2. In this case, the company should generate", + "references": "" + }, + { + "question": "A company operates large cranes at a busy port. The company plans to use machine learning (ML) for predictive maintenance of the cranes to avoid unexp ected breakdowns and to improve productivity. The company already uses sensor data from each cran e to monitor the health of the cranes in real time. The sensor data includes rotation speed, tens ion, energy consumption, vibration, pressure, and \u00a6perature for each crane. The company contracts AWS ML experts to implement an ML solution. Which potential findings would indicate that an ML- based solution is suitable for this scenario? (Select TWO.)", + "options": [ + "A. The historical sensor data does not include a sig nificant number of data points and attributes for", + "B. The historical sensor data shows that simple rule -based thresholds can predict crane failures.", + "C. The historical sensor data contains failure data for only one type of crane model that is in", + "D. The historical sensor data from the cranes are av ailable with high granularity for the last 3 years." + ], + "correct": "", + "explanation": "The best indicators that an ML-based solution is su itable for this scenario are D and E, because they imply that the historical sensor data is sufficient and relevant for building a predictive maintenancemodel. This model can use machine learning techniqu es such as regression, classification, or anomaly detection to learn from the past data and f orecast future failures or issues12. Having high granularity and diversity of data can improve the a ccuracy and generalization of the model, as well as enable the detection of complex patterns and relati onships that are not captured by simple rulebased thresholds3. The other options are not good indicators that an M L-based solution is suitable, because they suggest that the historical sensor data is incomple te, inconsistent, or inadequate for building a predictive maintenance model. These options would r equire additional data collection, preprocessing, or augmentation to overcome the data quality issues and ensure that the model can handle different scenarios and types of cranes4 .", + "references": "1: Machine Learning Techniques for Predictive Maint enance 2: A Guide to Predictive Maintenance & Machine Lear ning 3: Machine Learning for Predictive Maintenance: Rei nventing Asset Upkeep 4: Predictive Maintenance with Machine Learning: A Complete Guide : [Machine Learning for Predictive Maintenance - AW S Online Tech Talks]" + }, + { + "question": "A company wants to create an artificial intelligenc e (Al) yoga instructor that can lead large classes of students. The company needs to create a feature tha t can accurately count the number of students who are in a class. The company also needs a featur e that can differentiate students who are performing a yoga stretch correctly from students w ho are performing a stretch incorrectly. ...etermine whether students are performing a stret ch correctly, the solution needs to measure the location and angle of each student's arms and legs A data scientist must use Amazon SageMaker to ...ss video footage of a yoga class by extracting i mage frames and applying computer vision models. Which combination of models will meet these require ments with the LEAST effort? (Select TWO.)", + "options": [ + "A. Image Classification", + "B. Optical Character Recognition (OCR)", + "C. Object Detection", + "D. Pose estimation" + ], + "correct": "", + "explanation": "To count the number of students who are in a class, the solution needs to detect and locate each student in the video frame. Object detection is a c omputer vision model that can identify and locate multiple objects in an image. To differentiate stud ents who are performing a stretch correctly from students who are performing a stretch incorrectly, the solution needs to measure the location and angle of each students arms and legs. Pose estimati on is a computer vision model that can estimate the pose of a person by detecting the position and orientation of key body parts. Image classification , OCR, and image GANs are not relevant for this use c ase. Reference: Object Detection: A computer vision technique that identifies and locates objects within an image or video. Pose Estimation: A computer vision technique that e stimates the pose of a person by detecting the position and orientation of key body parts. Amazon SageMaker: A fully managed service that prov ides every developer and data scientist with the ability to build, train, and deploy machine lea rning (ML) models quickly.", + "references": "" + }, + { + "question": "A wildlife research company has a set of images of lions and cheetahs. The company created a dataset of the images. The company labeled each ima ge with a binary label that indicates whether an image contains a lion or cheetah. The company wa nts to train a model to identify whether new images contain a lion or cheetah. .... Dh Amazon SageMaker algorithm will meet this r equirement? A. XGBoost", + "options": [ + "B. Image Classification - TensorFlow", + "C. Object Detection - TensorFlow", + "D. Semantic segmentation - MXNet" + ], + "correct": "B. Image Classification - TensorFlow", + "explanation": "The best Amazon SageMaker algorithm for this task i s Image Classification - TensorFlow. This algorithm is a supervised learning algorithm that s upports transfer learning with many pretrained models from the TensorFlow Hub. Transfer learning a llows the company to fine-tune one of the available pretrained models on their own dataset, e ven if a large amount of image data is not available. The image classification algorithm takes an image as input and outputs a probability for each provided class label. The company can choose f rom a variety of models, such as MobileNet, ResNet, or Inception, depending on their accuracy a nd speed requirements. The algorithm also supports distributed training, data augmentation, a nd hyperparameter tuning.", + "references": "Image Classification - TensorFlow - Amazon SageMake r Amazon SageMaker Provides New Built-in TensorFlow I mage Classification Algorithm Image Classification with ResNet :: Amazon SageMake r Workshop Image classification on Amazon SageMaker | by Julie n Simon - Medium" + }, + { + "question": "An ecommerce company has used Amazon SageMaker to d eploy a factorization machines (FM) model to suggest products for customers. The compan y's data science team has developed two new models by using the TensorFlow and PyTorch deep lea rning frameworks. The company needs to use A/B testing to evaluate the new models against the deployed model. ...required A/B testing setup is as follows: Send 70% of traffic to the FM model, 15% of traffic to the TensorFlow model, and 15% of traffic to the Py Torch model. For customers who are from Europe, send all traffic to the TensorFlow model ..sh architecture can the company use to implement the required A/B testing setup?", + "options": [ + "A. Create two new SageMaker endpoints for the Tensor Flow and PyTorch models in addition to the", + "B. Create two production variants for the TensorFlow and PyTorch models. Create an auto scaling", + "C. Create two new SageMaker endpoints for the Tensor Flow and PyTorch models in addition to the", + "D. Create two production variants for the TensorFlow and PyTorch models. Specify the weight for" + ], + "correct": "", + "explanation": "The correct answer is D because it allows the compa ny to use the existing SageMaker endpoint and leverage the built-in functionality of production v ariants for A/B testing. Production variants can be used to test ML models that have been trained using different training datasets, algorithms, and ML frameworks; test how they perform on different inst ance types; or a combination of all of the above1. By specifying the weight for each productio n variant in the endpoint configuration, the company can control how much traffic to send to eac h variant. By setting the TargetVariant header in the request, the company can invoke a specific vari ant directly for each request2. This enables the company to implement the required A/B testing setup without creating additional endpoints or load balancers.", + "references": "1: Production variants - Amazon SageMaker 2: A/B Testing ML models in production using Amazon SageMaker | AWS Machine Learning Blog" + }, + { + "question": "A data scientist stores financial datasets in Amazo n S3. The data scientist uses Amazon Athena to query the datasets by using SQL. The data scientist uses Amazon SageMaker to deploy a machine learning (ML) model. The data scientist wants to obtain inferences from the model at the SageMaker endpoint However, when the data \u00a6. ntist attempts to invoke the SageMaker endp oint, the data scientist receives SOL statement failures The data scientist's 1AM user is currently unable to invoke the SageMaker endpoint Which combination of actions will give the data sci entist's 1AM user the ability to invoke the SageMaker endpoint? (Select THREE.)", + "options": [ + "A. Attach the AmazonAthenaFullAccess AWS managed pol icy to the user identity.", + "B. Include a policy statement for the data scientist 's 1AM user that allows the 1AM user to perform", + "C. Include an inline policy for the data scientists 1AM user that allows SageMaker to read S3 objects", + "D. Include a policy statement for the data scientist 's 1AM user that allows the 1AM user to perform" + ], + "correct": "", + "explanation": "The correct combination of actions to enable the da ta scientists IAM user to invoke the SageMaker endpoint is B, C, and E, because they ensure that t he IAM user has the necessary permissions, access, and syntax to query the ML model from Athen a. These actions have the following benefits: B: Including a policy statement for the IAM user th at allows the sagemaker:InvokeEndpoint action grants the IAM user the permission to call the Sage Maker Runtime InvokeEndpoint API, which is used to get inferences from the model hosted at the endpoint1. C: Including an inline policy for the IAM user that allows SageMaker to read S3 objects enables the IAM user to access the data stored in S3, which is the source of the Athena queries2. E: Including the SQL statement oeUSING EXTERNAL FUN CTION ml_function_name in the Athena SQL query allows the IAM user to invoke the ML model as an external function from Athena, which is a feature that enables querying ML models from SQL st atements3. The other options are not correct or necessary, bec ause they have the following drawbacks: A: Attaching the AmazonAthenaFullAccess AWS managed policy to the user identity is not sufficient, because it does not grant the IAM user the permissi on to invoke the SageMaker endpoint, which is required to query the ML model4. D: Including a policy statement for the IAM user th at allows the IAM user to perform the sagemaker:GetRecord action is not relevant, because this action is used to retrieve a single record from a feature group, which is not the case in this scenario5. F: Performing a user remapping in SageMaker to map the IAM user to another IAM user that is on the hosted endpoint is not applicable, because this feature is only available for multi-model endpoints, which are not used in this scenario.", + "references": "1: InvokeEndpoint - Amazon SageMaker 2: Querying Data in Amazon S3 from Amazon Athena - Amazon Athena 3: Querying machine learning models from Amazon Ath ena using Amazon SageMaker | AWS Machine Learning Blog 4: AmazonAthenaFullAccess - AWS Identity and Access Management 5: GetRecord - Amazon SageMaker Feature Store Runti me : [Invoke a Multi-Model Endpoint - Amazon SageMaker ]" + }, + { + "question": "A company is using Amazon SageMaker to build a mach ine learning (ML) model to predict customer churn based on customer call transcripts. Audio fil es from customer calls are located in an onpremises VoIP system that has petabytes of recorded calls. T he on-premises infrastructure has highvelocity networking and connects to the company's AWS infras tructure through a VPN connection over a 100 Mbps connection. The company has an algorithm for transcribing custo mer calls that requires GPUs for inference. The company wants to store these transcriptions in an A mazon S3 bucket in the AWS Cloud for model development. Which solution should an ML specialist use to deliv er the transcriptions to the S3 bucket as quickly a s possible?", + "options": [ + "A. Order and use an AWS Snowball Edge Compute Optimi zed device with an NVIDIA Tesla module to", + "B. Order and use an AWS Snowcone device with Amazon EC2 Inf1 instances to run the transcription", + "C. Order and use AWS Outposts to run the transcripti on algorithm on GPU-based Amazon EC2", + "D. Use AWS DataSync to ingest the audio files to Ama zon S3. Create an AWS Lambda function to run" + ], + "correct": "A. Order and use an AWS Snowball Edge Compute Optimi zed device with an NVIDIA Tesla module to", + "explanation": "The company needs to transcribe petabytes of audio files from an on-premises VoIP system to an S3 bucket in the AWS Cloud. The transcription algorith m requires GPUs for inference, which are not available on the on-premises system. The VPN connec tion over a 100 Mbps connection is not sufficient to transfer the large amount of data qui ckly. Therefore, the company should use an AWS Snowball Edge Compute Optimized device with an NVID IA Tesla module to run the transcription algorithm locally and leverage the GPU power. The d evice can store up to 42 TB of data and can be shipped back to AWS for data ingestion. The company can use AWS DataSync to send the resulting transcriptions to the transcription S3 bucket in th e AWS Cloud. This solution minimizes the network bandwidth and latency issues and enables faster dat a processing and transfer. Option B is incorrect because AWS Snowcone is a sma ll, portable, rugged, and secure edge computing and data transfer device that can store u p to 8 TB of data. It is not suitable for processin g petabytes of data and does not support GPU-based in stances. Option C is incorrect because AWS Outposts is a ser vice that extends AWS infrastructure, services, APIs, and tools to virtually any data center, co-lo cation space, or on-premises facility. It is not designed for data transfer and ingestion, and it wo uld require additional infrastructure and maintenance costs. Option D is incorrect because AWS DataSync is a ser vice that makes it easy to move large amounts of data to and from AWS over the internet or AWS Direc t Connect. However, using DataSync to ingest the audio files to S3 would still be limited by the network bandwidth and latency. Moreover, running the transcription algorithm on AWS Lambda would inc ur additional costs and complexity, and it would not leverage the GPU power that the algorithm requires.", + "references": "AWS Snowball Edge Compute Optimized AWS DataSync AWS Snowcone AWS Outposts AWS Lambda" + }, + { + "question": "A data scientist is building a linear regression mo del. The scientist inspects the dataset and notices that the mode of the distribution is lower than the median, and the median is lower than the mean. Which data transformation will give the data scient ist the ability to apply a linear regression model?", + "options": [ + "A. Exponential transformation", + "B. Logarithmic transformation", + "C. Polynomial transformation", + "D. Sinusoidal transformation" + ], + "correct": "B. Logarithmic transformation", + "explanation": "A logarithmic transformation is a suitable data tra nsformation for a linear regression model when the data has a skewed distribution, such as when the mo de is lower than the median and the median is lower than the mean. A logarithmic transformation c an reduce the skewness and make the data more symmetric and normally distributed, which are desirable properties for linear regression. A logarithmic transformation can also reduce the effe ct of outliers and heteroscedasticity (unequal variance) in the data. An exponential transformatio n would have the opposite effect of increasing the skewness and making the data more asymmetric. A pol ynomial transformation may not be able to capture the nonlinearity in the data and may introd uce multicollinearity among the transformed variables. A sinusoidal transformation is not appro priate for data that does not have a periodic pattern.", + "references": "Data Transformation - Scaler Topics Linear Regression - GeeksforGeeks Linear Regression - Scribbr" + }, + { + "question": "A company is planning a marketing campaign to promo te a new product to existing customers. The company has data (or past promotions that are simil ar. The company decides to try an experiment to send a more expensive marketing package to a smalle r number of customers. The company wants to target the marketing campaign to customers who are most likely to buy the new product. The experiment requires that at least 90% of the custom ers who are likely to purchase the new product receive the marketing materials. ...company trains a model by using the linear learn er algorithm in Amazon SageMaker. The model has a recall score of 80% and a precision of 75%. ...should the company retrain the model to meet the se requirements?", + "options": [ + "A. Set the target_recall hyperparameter to 90% Set t he binaryclassrfier model_selection_critena hyperparameter to recall_at_target_precision.", + "B. Set the targetprecision hyperparameter to 90%. Se t the binary classifier model selection criteria", + "C. Use 90% of the historical data for training Set t he number of epochs to 20.", + "D. Set the normalize_jabel hyperparameter to true. S et the number of classes to 2." + ], + "correct": "A. Set the target_recall hyperparameter to 90% Set t he binaryclassrfier model_selection_critena hyperparameter to recall_at_target_precision.", + "explanation": "The best way to retrain the model to meet the requi rements is to set the target_recall hyperparameter to 90% and set the binary_classifier _model_selection_criteria hyperparameter to recall_at_target_precision. This will instruct the linear learner algorithm to optimize the model for a high recall score, while maintaining a reasonable p recision score. Recall is the proportion of actual positives that were identified correctly, which is important for the companys goal of reaching at leas t 90% of the customers who are likely to buy the new product1. Precision is the proportion of positive identifications that were actually correct, which i s also relevant for the companys budget and efficiency2. By setting the target_recall to 90%, t he algorithm will try to achieve a recall score of at least 90%, and by setting the binary_classifier_mod el_selection_criteria to recall_at_target_precision, the algorithm will sele ct the model that has the highest recall score among those that have a precision score equal to or higher than the target precision3. The target precision is automatically set to the median of the precision scores of all the models trained in parallel4. The other options are not correct or optimal, becau se they have the following drawbacks: B: Setting the target_precision hyperparameter to 9 0% and setting the binary_classifier_model_selection_criteria hyperpar ameter to precision_at_target_recall will optimize the model for a high precision score, whil e maintaining a reasonable recall score. However, this is not aligned with the companys goal of reach ing at least 90% of the customers who are likely to buy the new product, as precision does not refle ct how well the model identifies the actual positives1. Moreover, setting the target_precision to 90% might be too high and unrealistic for the dataset, as the current precision score is only 75% 4. C: Using 90% of the historical data for training an d setting the number of epochs to 20 will not necessarily improve the recall score of the model, as it does not change the optimization objective or the model selection criteria. Moreover, using more data for training might reduce the amount of data available for validation, which is needed for selec ting the best model among the ones trained in parallel3. The number of epochs is also not a decis ive factor for the recall score, as it depends on t he learning rate, the optimizer, and the convergence o f the algorithm5. D: Setting the normalize_label hyperparameter to tr ue and setting the number of classes to 2 will not affect the recall score of the model, as these are irrelevant hyperparameters for binary classificatio n problems. The normalize_label hyperparameter is onl y applicable for regression problems, as it controls whether the label is normalized to have ze ro mean and unit variance3. The number of classes hyperparameter is only applicable for multi class classification problems, as it specifies the number of output classes3.", + "references": "1: Classification: Precision and Recall | Machine L earning | Google for Developers 2: Precision and recall - Wikipedia 3: Linear Learner Algorithm - Amazon SageMaker 4: How linear learner works - Amazon SageMaker 5: Getting hands-on with Amazon SageMaker Linear Le arner - Pluralsight" + }, + { + "question": "A data scientist receives a collection of insurance claim records. Each record includes a claim ID. th e final outcome of the insurance claim, and the date of the final outcome. The final outcome of each claim is a selection from among 200 outcome categories. Some claim records include only partial information. However, incomplete claim records include only 3 or 4 outcome ...gones from among the 200 available outco me categories. The collection includes hundreds of records for each outcome category. The records are from the previous 3 years. The data scientist must create a solution to predic t the number of claims that will be in each outcome category every month, several months in advance. Which solution will meet these requirements?", + "options": [ + "A. Perform classification every month by using super vised learning of the 20X3 outcome categories", + "B. Perform reinforcement learning by using claim IDs and dates Instruct the insurance agents who", + "C. Perform forecasting by using claim IDs and dates to identify the expected number ot claims in each", + "D. Perform classification by using supervised learni ng of the outcome categories for which partial" + ], + "correct": "C. Perform forecasting by using claim IDs and dates to identify the expected number ot claims in each", + "explanation": "The best solution for this scenario is to perform f orecasting by using claim IDs and dates to identify the expected number of claims in each outcome categ ory every month. This solution has the following advantages: It leverages the historical data of claim outcomes and dates to capture the temporal patterns and trends of the claims in each category1. It does not require the claim contents or any other features to make predictions, which simplifies the data preparation and reduces the impact of missing or incomplete data2. It can handle the high cardinality of the outcome c ategories, as forecasting models can output multiple values for each time point3. It can provide predictions for several months in ad vance, which is useful for planning and budgeting purposes4. The other solutions have the following drawbacks: A: Performing classification every month by using s upervised learning of the 200 outcome categories based on claim contents is not suitable, because it assumes that the claim contents are available and complete for all the records, which is not the case in this scenario2. Moreover, classification models usually output a single label for each input, which is not adequate for predicting the number of claims in each category3. Additionally, classificat ion models do not account for the temporal aspect of the data, which is important for forecasting1. B: Performing reinforcement learning by using claim IDs and dates and instructing the insurance agents who submit the claim records to estimate the expected number of claims in each outcome category every month is not feasible, because it re quires a feedback loop between the model and the agents, which might not be available or reliabl e in this scenario5. Furthermore, reinforcement learning is more suitable for sequential decision m aking problems, where the model learns from its actions and rewards, rather than forecasting proble ms, where the model learns from historical data and outputs future values6. D: Performing classification by using supervised le arning of the outcome categories for which partial information on claim contents is provided and perfo rming forecasting by using claim IDs and dates for all other outcome categories is not optimal, be cause it combines two different methods that might not be consistent or compatible with each oth er7. Also, this solution suffers from the same limitations as solution A, such as the dependency o n claim contents, the inability to handle multiple outputs, and the ignorance of temporal patterns123.", + "references": "1: Time Series Forecasting - Amazon SageMaker 2: Handling Missing Data for Machine Learning | AWS Machine Learning Blog 3: Forecasting vs Classification: Whats the Differe nce? | DataRobot 4: Amazon Forecast \" Time Series Forecasting Made E asy | AWS News Blog 5: Reinforcement Learning - Amazon SageMaker 6: What is Reinforcement Learning? The Complete Gui de | Edureka 7: Combining Machine Learning Models | by Will Koeh rsen | Towards Data Science" + }, + { + "question": "A retail company stores 100 GB of daily transaction al data in Amazon S3 at periodic intervals. The company wants to identify the schema of the transac tional dat", + "options": [ + "A. The company also wants to perform transformations on the transactional data that is in Amazon", + "B. Use Amazon Athena to scan the data and identify t he schema.", + "C. Use AWS Glue crawlers to scan the data and identi fy the schema.", + "D. Use Amazon Redshift to store procedures to perfor m data transformations" + ], + "correct": "", + "explanation": "To meet the requirements with the least operational overhead, the company should use AWS Glue crawlers, AWS Glue workflows and jobs, and Amazon F raud Detector. AWS Glue crawlers can scan the data in Amazon S3 and identify the schema, whic h is then stored in the AWS Glue Data Catalog. AWS Glue workflows and jobs can perform data transf ormations on the data in Amazon S3 using serverless Spark or Python scripts. Amazon Fraud De tector can train a model to detect fraud using the transformed data and the companys historical fr aud labels, and then generate fraud predictions using a simple API call. Option A is incorrect because Amazon Athena is a se rverless query service that can analyze data in Amazon S3 using standard SQL, but it does not perfo rm data transformations or fraud detection. Option C is incorrect because Amazon Redshift is a cloud data warehouse that can store and query data using SQL, but it requires provisioning and ma naging clusters, which adds operational overhead. Moreover, Amazon Redshift does not provide a built- in fraud detection capability. Option E is incorrect because Amazon Redshift ML is a feature that allows users to create, train, and deploy machine learning models using SQL commands i n Amazon Redshift. However, using Amazon Redshift ML would require loading the data from Ama zon S3 to Amazon Redshift, which adds complexity and cost. Also, Amazon Redshift ML does not support fraud detection as a use case.", + "references": "AWS Glue Crawlers AWS Glue Workflows and Jobs Amazon Fraud Detector" + }, + { + "question": "A data scientist uses Amazon SageMaker Data Wrangle r to define and perform transformations and feature engineering on historical dat", + "options": [ + "A. The data scientist saves the transformations to S ageMaker Feature Store.", + "B. Use AWS Lambda to run a predefined SageMaker pipe line to perform the transformations on each", + "D. Use Apache Airflow to orchestrate a set of predef ined transformations on each new dataset that" + ], + "correct": "D. Use Apache Airflow to orchestrate a set of predef ined transformations on each new dataset that", + "explanation": "The best solution is to configure Amazon EventBridg e to run a predefined SageMaker pipeline to perform the transformations when a new data is dete cted in the S3 bucket. This solution requires the least development effort because it leverages the n ative integration between EventBridge and SageMaker Pipelines, which allows you to trigger a pipeline execution based on an event rule. EventBridge can monitor the S3 bucket for new data uploads and invoke the pipeline that contains the same transformations and feature engineering st eps that were defined in SageMaker Data Wrangler. The pipeline can then ingest the transfor med data into the online feature store for training and inference. The other solutions are less optimal because they r equire more development effort and additional services. Using AWS Lambda or AWS Step Functions wo uld require writing custom code to invoke the SageMaker pipeline and handle any errors or retries . Using Apache Airflow would require setting up and maintaining an Airflow server and DAGs, as well as integrating with the SageMaker API.", + "references": "Amazon EventBridge and Amazon SageMaker Pipelines i ntegration Create a pipeline using a JSON specification Ingest data into a feature group" + }, + { + "question": "A data scientist at a financial services company us ed Amazon SageMaker to train and deploy a model that predicts loan defaults. The model analyzes new loan applications and predicts the risk of loan default. To train the model, the data scientist man ually extracted loan data from a database. The data scientist performed the model training and deployme nt steps in a Jupyter notebook that is hosted on SageMaker Studio notebooks. The model's prediction accuracy is decreasing over time. Which combination of slept in the MOST operationall y efficient way for the data scientist to maintain the model's accuracy? (Select TWO.)", + "options": [ + "A. Use SageMaker Pipelines to create an automated wo rkflow that extracts fresh data, trains the", + "B. Configure SageMaker Model Monitor with an accurac y threshold to check for model drift. Initiate", + "C. Store the model predictions in Amazon S3 Create a daily SageMaker Processing job that reads the", + "D. Rerun the steps in the Jupyter notebook that is h osted on SageMaker Studio notebooks to retrain" + ], + "correct": "", + "explanation": "Explanation/Reference: Option A is correct because SageMaker Pipelines is a service that enables you to create and manage automated workflows for your machine learning proje cts. You can use SageMaker Pipelines to orchestrate the steps of data extraction, model tra ining, and model deployment in a repeatable and scalable way1. Option B is correct because SageMaker Model Monitor is a service that monitors the quality of your models in production and alerts you when there are deviations in the model quality. You can use SageMaker Model Monitor to set an accuracy threshol d for your model and configure a CloudWatch alarm that triggers when the threshold is exceeded. You can then connect the alarm to the workflow in SageMaker Pipelines to automatically initiate re training and deployment of a new version of the model2. Option C is incorrect because it is not the most op erationally efficient way to maintain the models accuracy. Creating a daily SageMaker Processing job that reads the predictions from Amazon S3 and checks for changes in model prediction accuracy is a manual and time-consuming process. It also requires you to write custom code to perform the da ta analysis and send the email notification. Moreover, it does not automatically retrain and dep loy the model when the accuracy drops. Option D is incorrect because it is not the most op erationally efficient way to maintain the models accuracy. Rerunning the steps in the Jupyter notebo ok that is hosted on SageMaker Studio notebooks to retrain the model and redeploy a new version of the model is a manual and error-prone process. It also requires you to monitor the models performance and initiate the retraining and deployment steps yourself. Moreover, it does not leverage the benefits of SageMaker Pipelines and SageMaker Model Monitor to automate and streamline the workfl ow. Option E is incorrect because it is not the most op erationally efficient way to maintain the models accuracy. Exporting the training and deployment cod e from the SageMaker Studio notebooks into a Python script and packaging the script into an Amaz on ECS task that an AWS Lambda function can initiate is a complex and cumbersome process. It al so requires you to manage the infrastructure and resources for the Amazon ECS task and the AWS Lambd a function. Moreover, it does not leverage the benefits of SageMaker Pipelines and SageMaker Model Monitor to automate and streamline the workflow.", + "references": "1: SageMaker Pipelines - Amazon SageMaker 2: Monitor data and model quality - Amazon SageMake r" + }, + { + "question": "An insurance company developed a new experimental m achine learning (ML) model to replace an existing model that is in production. The company m ust validate the quality of predictions from the new experimental model in a production environment before the company uses the new experimental model to serve general user requests. Which one model can serve user requests at a time. The company must measure the performance of the new experimental model without affecting the cu rrent live traffic Which solution will meet these requirements?", + "options": [ + "A. A/B testing", + "B. Canary release", + "C. Shadow deployment", + "D. Blue/green deployment" + ], + "correct": "C. Shadow deployment", + "explanation": "The best solution for this scenario is to use shado w deployment, which is a technique that allows the company to run the new experimental model in parall el with the existing model, without exposing it to the end users. In shadow deployment, the company can route the same user requests to both models, but only return the responses from the exis ting model to the users. The responses from the new experimental model are logged and analyzed for quality and performance metrics, such as accuracy, latency, and resource consumption12. This way, the company can validate the new experimental model in a production environment, wit hout affecting the current live traffic or user experience. The other solutions are not suitable, because they have the following drawbacks: A: A/B testing is a technique that involves splitti ng the user traffic between two or more models, and comparing their outcomes based on predefined metric s. However, this technique exposes the new experimental model to a portion of the end users, w hich might affect their experience if the model is not reliable or consistent with the existing model3 . B: Canary release is a technique that involves grad ually rolling out the new experimental model to a small subset of users, and monitoring its performan ce and feedback. However, this technique also exposes the new experimental model to some end user s, and requires careful selection and segmentation of the user groups4. D: Blue/green deployment is a technique that involv es switching the user traffic from the existing model (blue) to the new experimental model (green) at once, after testing and verifying the new model in a separate environment. However, this tech nique does not allow the company to validate the new experimental model in a production environm ent, and might cause service disruption or inconsistency if the new model is not compatible or stable5.", + "references": "1: Shadow Deployment: A Safe Way to Test in Product ion | LaunchDarkly Blog 2: Shadow Deployment: A Safe Way to Test in Product ion | LaunchDarkly Blog 3: A/B Testing for Machine Learning Models | AWS Ma chine Learning Blog 4: Canary Releases for Machine Learning Models | AW S Machine Learning Blog 5: Blue-Green Deployments for Machine Learning Mode ls | AWS Machine Learning Blog" + }, + { + "question": "An ecommerce company wants to use machine learning (ML) to monitor fraudulent transactions on its website. The company is using Amazon SageMaker to research, train, deploy, and monitor the ML models. The historical transactions data is in a .csv file that is stored in Amazon S3 The data contains featu res such as the user's IP address, navigation time, ave rage time on each page, and the number of clicks for ....session. There is no label in the data to i ndicate if a transaction is anomalous. Which models should the company use in combination to detect anomalous transactions? (Select TWO.)", + "options": [ + "A. IP Insights", + "B. K-nearest neighbors (k-NN)", + "C. Linear learner with a logistic function", + "D. Random Cut Forest (RCF)" + ], + "correct": "", + "explanation": "To detect anomalous transactions, the company can u se a combination of Random Cut Forest (RCF) and XGBoost models. RCF is an unsupervised algorith m that can detect outliers in the data by measuring the depth of each data point in a collect ion of random decision trees. XGBoost is a supervised algorithm that can learn from the labele d data points generated by RCF and classify them as normal or anomalous. RCF can also provide anomal y scores that can be used as features for XGBoost to improve the accuracy of the classificati on. Reference: 1: Amazon SageMaker Random Cut Forest 2: Amazon SageMaker XGBoost Algorithm 3: Anomaly Detection with Amazon SageMaker Random C ut Forest and Amazon SageMaker XGBoost", + "references": "" + }, + { + "question": "A finance company needs to forecast the price of a commodity. The company has compiled a dataset of historical daily prices. A data scientist must t rain various forecasting models on 80% of the datas et and must validate the efficacy of those models on t he remaining 20% of the dataset. What should the data scientist split the dataset in to a training dataset and a validation dataset to compare model performance?", + "options": [ + "A. Pick a date so that 80% to the data points preced e the date Assign that group of data points as the", + "B. Pick a date so that 80% of the data points occur after the date. Assign that group of data points as", + "C. Starting from the earliest date in the dataset. p ick eight data points for the training dataset and", + "D. Sample data points randomly without replacement s o that 80% of the data points are in the" + ], + "correct": "A. Pick a date so that 80% to the data points preced e the date Assign that group of data points as the", + "explanation": "A Comprehensive The best way to split the dataset i nto a training dataset and a validation dataset is to pick a date so that 80% of the data points prece de the date and assign that group of data points as the training dataset. This method preserves the tem poral order of the data and ensures that the validation dataset reflects the most recent trends and patterns in the commodity price. This is important for forecasting models that rely on time series analysis and sequential data. The other methods would either introduce bias or lose informa tion by ignoring the temporal structure of the data.", + "references": "Time Series Forecasting - Amazon SageMaker Time Series Splitting - scikit-learn Time Series Forecasting - Towards Data Science" + }, + { + "question": "A manufacturing company needs to identify returned smartphones that have been damaged by moisture. The company has an automated process that produces 2.000 diagnostic values for each phone. The database contains more than five million phone evaluations. The evaluation process is consistent, and there are no missing values in the dat", + "options": [ + "A. A machine learning (ML) specialist has trained an Amazon SageMaker linear learner ML model to", + "B. Continue to use the SageMaker linear learner algo rithm. Reduce the number of features with the", + "C. Continue to use the SageMaker linear learner algo rithm. Reduce the number of features with the", + "D. Continue to use the SageMaker linear learner algo rithm. Set the predictor type to regressor." + ], + "correct": "", + "explanation": "Option A is correct because reducing the number of features with the SageMaker PCA algorithm can help remove noise and redundancy from the data, and improve the models performance. PCA is a dimensionality reduction technique that transforms the original features into a smaller set of linearl y uncorrelated features called principal components. The SageMaker linear learner algorithm supports PCA as a built-in feature transformation option. Option E is correct because using the SageMaker k-N N algorithm with a dimension reduction target of less than 1,000 can help the model learn from th e similarity of the data points, and improve the models performance. k-NN is a non-parametric algori thm that classifies an input based on the majority vote of its k nearest neighbors in the fea ture space. The SageMaker k-NN algorithm supports dimension reduction as a built-in feature transformation option. Option B is incorrect because using the scikit-lear n MDS algorithm to reduce the number of features is not a feasible option, as MDS is a computational ly expensive technique that does not scale well to large datasets. MDS is a dimensionality reduction t echnique that tries to preserve the pairwise distances between the original data points in a low er-dimensional space. Option C is incorrect because setting the predictor type to regressor would change the models objective from classification to regression, which is not suitable for the given problem. A regressor model would output a continuous value instead of a binary label for each phone. Option D is incorrect because using the SageMaker k -means algorithm with k of less than 1,000 would not help the model classify the phones, as k- means is a clustering algorithm that groups the data points into k clusters based on their similari ty, without using any labels. A clustering model would not output a binary label for each phone.", + "references": "Amazon SageMaker Linear Learner Algorithm Amazon SageMaker K-Nearest Neighbors (k-NN) Algorit hm [Principal Component Analysis - Scikit-learn] [Multidimensional Scaling - Scikit-learn]" + }, + { + "question": "A company deployed a machine learning (ML) model on the company website to predict real estate prices. Several months after deployment, an ML engi neer notices that the accuracy of the model has gradually decreased. The ML engineer needs to improve the accuracy of th e model. The engineer also needs to receive notifications for any future performance issues. Which solution will meet these requirements?", + "options": [ + "A. Perform incremental training to update the model. Activate Amazon SageMaker Model Monitor to", + "B. Use Amazon SageMaker Model Governance. Configure Model Governance to automatically adjust", + "C. Use Amazon SageMaker Debugger with appropriate th resholds. Configure Debugger to send", + "D. Use only data from the previous several months to perform incremental training to update the" + ], + "correct": "A. Perform incremental training to update the model. Activate Amazon SageMaker Model Monitor to", + "explanation": "The best solution to improve the accuracy of the mo del and receive notifications for any future performance issues is to perform incremental traini ng to update the model and activate Amazon SageMaker Model Monitor to detect model performance issues and to send notifications. Incremental training is a technique that allows you to update an existing model with new data without retraining the entire model from scratch. T his can save time and resources, and help the model adapt to changing data patterns. Amazon SageM aker Model Monitor is a feature that continuously monitors the quality of machine learni ng models in production and notifies you when there are deviations in the model quality, such as data drift and anomalies. You can set up alerts tha t trigger actions, such as sending notifications to A mazon Simple Notification Service (Amazon SNS) topics, when certain conditions are met. Option B is incorrect because Amazon SageMaker Mode l Governance is a set of tools that help you implement ML responsibly by simplifying access cont rol and enhancing transparency. It does not provide a mechanism to automatically adjust model h yperparameters or improve model accuracy. Option C is incorrect because Amazon SageMaker Debu gger is a feature that helps you debug and optimize your model training process by capturing r elevant data and providing real-time analysis. However, using Debugger alone does not update the m odel or monitor its performance in production. Also, retraining the model by using onl y data from the previous several months may not capture the full range of data variability and may introduce bias or overfitting. Option D is incorrect because using only data from the previous several months to perform incremental training may not be sufficient to impro ve the model accuracy, as explained above. Moreover, this option does not specify how to activ ate Amazon SageMaker Model Monitor or configure the alerts and notifications.", + "references": "Incremental training Amazon SageMaker Model Monitor Amazon SageMaker Model Governance Amazon SageMaker Debugger" + }, + { + "question": "A university wants to develop a targeted recruitmen t strategy to increase new student enrollment. A data scientist gathers information about the academ ic performance history of students. The data scientist wants to use the data to build student pr ofiles. The university will use the profiles to dir ect resources to recruit students who are likely to enr oll in the university. Which combination of steps should the data scientis t take to predict whether a particular student applicant is likely to enroll in the university? (S elect TWO)", + "options": [ + "A. Use Amazon SageMaker Ground Truth to sort the dat a into two groups named \"enrolled\" or \"not", + "B. Use a forecasting algorithm to run predictions.", + "C. Use a regression algorithm to run predictions.", + "D. Use a classification algorithm to run predictions" + ], + "correct": "", + "explanation": "The data scientist should use Amazon SageMaker Grou nd Truth to sort the data into two groups named oeenrolled or oenot enrolled. This will creat e a labeled dataset that can be used for supervised learning. The data scientist should then use a clas sification algorithm to run predictions on the test data. A classification algorithm is a suitable choi ce for predicting a binary outcome, such as enrollment status, based on the input features, suc h as academic performance. A classification algorithm will output a probability for each class label and assign the most likely label to each observation.", + "references": "Use Amazon SageMaker Ground Truth to Label Data Classification Algorithm in Machine Learning" + }, + { + "question": "A company's machine learning (ML) specialist is bui lding a computer vision model to classify 10 different traffic signs. The company has stored 100 images of each class in Amazon S3, and the company has another 10.000 unlabeled images. All th e images come from dash cameras and are a size of 224 pixels * 224 pixels. After several trai ning runs, the model is overfitting on the training data. Which actions should the ML specialist take to addr ess this problem? (Select TWO.)", + "options": [ + "A. Use Amazon SageMaker Ground Truth to label the un labeled images", + "B. Use image preprocessing to transform the images i nto grayscale images.", + "C. Use data augmentation to rotate and translate the labeled images.", + "D. Replace the activation of the last layer with a s igmoid." + ], + "correct": "", + "explanation": "Data augmentation is a technique to increase the si ze and diversity of the training data by applying random transformations such as rotation, translatio n, scaling, flipping, etc. This can help reduce overfitting and improve the generalization of the m odel. Data augmentation can be done using the Amazon SageMaker image classification algorithm, wh ich supports various augmentation options such as horizontal_flip, vertical_flip, rotate, bri ghtness, contrast, etc1 The Amazon SageMaker k-nearest neighbors (k-NN) alg orithm is a supervised learning algorithm that can be used to label unlabeled data based on the si milarity to the labeled data. The k-NN algorithm assigns a label to an unlabeled instance by finding the k closest labeled instances in the feature spa ce and taking a majority vote among their labels. This can help increase the size and diversity of the training data and reduce overfitting. The k-NN algo rithm can be used with the Amazon SageMaker image classification algorithm by extracting featur es from the images using a pre-trained model and then applying the k-NN algorithm on the feature vec tors2 Using Amazon SageMaker Ground Truth to label the un labeled images is not a good option because it is a manual and costly process that requires hum an annotators. Moreover, it does not address the issue of overfitting on the existing labeled data. Using image preprocessing to transform the images i nto grayscale images is not a good option because it reduces the amount of information and va riation in the images, which can degrade the performance of the model. Moreover, it does not add ress the issue of overfitting on the existing labeled data. Replacing the activation of the last layer with a s igmoid is not a good option because it is not suita ble for a multi-class classification problem. A sigmoid activation function outputs a value between 0 and 1, which can be interpreted as a probability of bel onging to a single class. However, for a multi-clas s classification problem, the output should be a vect or of probabilities that sum up to 1, which can be achieved by using a softmax activation function.", + "references": "1: Image classification algorithm - Amazon SageMake r 2: k-nearest neighbors (k-NN) algorithm - Amazon Sa geMaker" + }, + { + "question": "A machine learning (ML) specialist is using the Ama zon SageMaker DeepAR forecasting algorithm to train a model on CPU-based Amazon EC2 On-Demand ins tances. The model currently takes multiple hours to train. The ML specialist wants to decrease the training time of the model. Which approaches will meet this requirement7 (SELEC T TWO )", + "options": [ + "A. Replace On-Demand Instances with Spot Instances", + "B. Configure model auto scaling dynamically to adjus t the number of instances automatically.", + "C. Replace CPU-based EC2 instances with GPU-based EC 2 instances.", + "D. Use multiple training instances." + ], + "correct": "", + "explanation": "The best approaches to decrease the training time o f the model are C and D, because they can improve the computational efficiency and paralleliz ation of the training process. These approaches have the following benefits: C: Replacing CPU-based EC2 instances with GPU-based EC2 instances can speed up the training of the DeepAR algorithm, as it can leverage the parallel p rocessing power of GPUs to perform matrix operations and gradient computations faster than CP Us12. The DeepAR algorithm supports GPUbased EC2 instances such as ml.p2 and ml.p33. D: Using multiple training instances can also reduc e the training time of the DeepAR algorithm, as it can distribute the workload across multiple nodes a nd perform data parallelism4. The DeepAR algorithm supports distributed training with multip le CPU-based or GPU-based EC2 instances3. The other options are not effective or relevant, be cause they have the following drawbacks: A: Replacing On-Demand Instances with Spot Instance s can reduce the cost of the training, but not necessarily the time, as Spot Instances are subject to interruption and availability5. Moreover, the DeepAR algorithm does not support checkpointing, wh ich means that the training cannot resume from the last saved state if the Spot Instance is t erminated3. B: Configuring model auto scaling dynamically to ad just the number of instances automatically is not applicable, as this feature is only available for i nference endpoints, not for training jobs6. E: Using a pre-trained version of the model and run ning incremental training is not possible, as the DeepAR algorithm does not support incremental train ing or transfer learning3. The DeepAR algorithm requires a full retraining of the model w henever new data is added or the hyperparameters are changed7.", + "references": "1: GPU vs CPU: What Matters Most for Machine Learni ng? | by Louis (Whats AI) Bouchard | Towards Data Science 2: How GPUs Accelerate Machine Learning Training | NVIDIA Developer Blog 3: DeepAR Forecasting Algorithm - Amazon SageMaker 4: Distributed Training - Amazon SageMaker 5: Managed Spot Training - Amazon SageMaker 6: Automatic Scaling - Amazon SageMaker 7: How the DeepAR Algorithm Works - Amazon SageMake r" + }, + { + "question": "An engraving company wants to automate its quality control process for plaques. The company performs the process before mailing each customized plaque to a customer. The company has created an Amazon S3 bucket that contains images of defects that should cause a plaque to be rejected. Low-confidence predictions must be sent t o an internal team of reviewers who are using Amazon Augmented Al (Amazon A2I). Which solution will meet these requirements?", + "options": [ + "A. Use Amazon Textract for automatic processing. Use Amazon A2I with Amazon Mechanical Turk for", + "B. Use Amazon Rekognition for automatic processing. Use Amazon A2I with a private workforce", + "C. Use Amazon Transcribe for automatic processing. U se Amazon A2I with a private workforce option", + "D. Use AWS Panorama for automatic processing Use Ama zon A2I with Amazon Mechanical Turk for" + ], + "correct": "B. Use Amazon Rekognition for automatic processing. Use Amazon A2I with a private workforce", + "explanation": "Amazon Rekognition is a service that provides compu ter vision capabilities for image and video analysis, such as object, scene, and activity detec tion, face and text recognition, and custom label detection. Amazon Rekognition can be used to automa te the quality control process for plaques by comparing the images of the plaques with the images of defects in the Amazon S3 bucket and returning a confidence score for each defect. Amazo n A2I is a service that enables human review of machine learning predictions, such as low-confidenc e predictions from Amazon Rekognition. Amazon A2I can be integrated with a private workforce opti on, which allows the engraving company to use its own internal team of reviewers to manually inspect the plaques that are flagged by Amazon Rekognition. This solution meets the requirements o f automating the quality control process, sending low-confidence predictions to an internal t eam of reviewers, and using Amazon A2I for manual review. Reference: 1: Amazon Rekognition documentation 2: Amazon A2I documentation 3: Amazon Rekognition Custom Labels documentation 4: Amazon A2I Private Workforce documentation", + "references": "" + }, + { + "question": "An online delivery company wants to choose the fast est courier for each delivery at the moment an order is placed. The company wants to implement thi s feature for existing users and new users of its application. Data scientists have trained separate models with XGBoost for this purpose, and the models are stored in Amazon S3. There is one model fof each city where the company operates. The engineers are hosting these models in Amazon EC 2 for responding to the web client requests, with one instance for each model, but the instances have only a 5% utilization in CPU and memory, ....operation engineers want to avoid managing unne cessary resources. Which solution will enable the company to achieve i ts goal with the LEAST operational overhead?", + "options": [ + "A. Create an Amazon SageMaker notebook instance for pulling all the models from Amazon S3 using", + "B. Prepare an Amazon SageMaker Docker container base d on the open-source multi-model server.", + "C. Keep only a single EC2 instance for hosting all t he models. Install a model server in the instance", + "D. Prepare a Docker container based on the prebuilt images in Amazon SageMaker. Replace the" + ], + "correct": "B. Prepare an Amazon SageMaker Docker container base d on the open-source multi-model server.", + "explanation": "The best solution for this scenario is to use a mul ti-model endpoint in Amazon SageMaker, which allows hosting multiple models on the same endpoint and invoking them dynamically at runtime. This way, the company can reduce the operational ov erhead of managing multiple EC2 instances and model servers, and leverage the scalability, securi ty, and performance of SageMaker hosting services. By using a multi-model endpoint, the comp any can also save on hosting costs by improving endpoint utilization and paying only for the models that are loaded in memory and the API calls that are made. To use a multi-model endpoint, the compan y needs to prepare a Docker container based on the open-source multi-model server, which is a f ramework-agnostic library that supports loading and serving multiple models from Amazon S3. The com pany can then create a multi-model endpoint in SageMaker, pointing to the S3 bucket containing all the models, and invoke the endpoint from the web client at runtime, specifying the TargetModel p arameter according to the city of each request. This solution also enables the company to add or re move models from the S3 bucket without redeploying the endpoint, and to use different vers ions of the same model for different cities if needed. Reference: Use Docker containers to build models Host multiple models in one container behind one en dpoint Multi-model endpoints using Scikit Learn Multi-model endpoints using XGBoost", + "references": "" + }, + { + "question": "A company builds computer-vision models that use de ep learning for the autonomous vehicle industry. A machine learning (ML) specialist uses a n Amazon EC2 instance that has a CPU: GPU ratio of 12:1 to train the models. The ML specialist examines the instance metric logs and notices that the GPU is idle half of the time The ML specialist must reduce training costs withou t increasing the duration of the training jobs. Which solution will meet these requirements?", + "options": [ + "A. Switch to an instance type that has only CPUs.", + "B. Use a heterogeneous cluster that has two differen t instances groups.", + "C. Use memory-optimized EC2 Spot Instances for the t raining jobs.", + "D. Switch to an instance type that has a CPU GPU rat io of 6:1." + ], + "correct": "D. Switch to an instance type that has a CPU GPU rat io of 6:1.", + "explanation": "Switching to an instance type that has a CPU: GPU r atio of 6:1 will reduce the training costs by using fewer CPUs and GPUs, while maintaining the same lev el of performance. The GPU idle time indicates that the CPU is not able to feed the GPU with enoug h data, so reducing the CPU: GPU ratio will balance the workload and improve the GPU utilizatio n. A lower CPU: GPU ratio also means less overhead for inter-process communication and synchr onization between the CPU and GPU processes. Reference: Optimizing GPU utilization for AI/ML workloads on A mazon EC2 Analyze CPU vs. GPU Performance for AWS Machine Lea rning", + "references": "" + }, + { + "question": "A company is building a new supervised classificati on model in an AWS environment. The company's data science team notices that the dataset has a la rge quantity of variables Ail the variables are numeric. The model accuracy for training and valida tion is low. The model's processing time is affected by high latency The data science team need s to increase the accuracy of the model and decrease the processing. How it should the data science team do to meet thes e requirements?", + "options": [ + "A. Create new features and interaction variables.", + "B. Use a principal component analysis (PCA) model.", + "C. Apply normalization on the feature set.", + "D. Use a multiple correspondence analysis (MCA) mode l" + ], + "correct": "B. Use a principal component analysis (PCA) model.", + "explanation": "The best way to meet the requirements is to use a p rincipal component analysis (PCA) model, which is a technique that reduces the dimensionality of t he dataset by transforming the original variables into a smaller set of new variables, called princip al components, that capture most of the variance and information in the data1. This technique has th e following advantages: It can increase the accuracy of the model by removi ng noise, redundancy, and multicollinearity from the data, and by enhancing the interpretability and generalization of the model23. It can decrease the processing time of the model by reducing the number of features and the computational complexity of the model, and by impro ving the convergence and stability of the model45. It is suitable for numeric variables, as it relies on the covariance or correlation matrix of the data , and it can handle a large quantity of variables, as it can extract the most relevant ones16. The other options are not effective or appropriate, because they have the following drawbacks: A: Creating new features and interaction variables can increase the accuracy of the model by capturing more complex and nonlinear relationships in the data, but it can also increase the processing time of the model by adding more feature s and increasing the computational complexity of the model7. Moreover, it can introduce more nois e, redundancy, and multicollinearity in the data, which can degrade the performance and interpretabil ity of the model8. C: Applying normalization on the feature set can in crease the accuracy of the model by scaling the features to a common range and avoiding the dominan ce of some features over others, but it can also decrease the processing time of the model by r educing the numerical instability and improving the convergence of the model . However, normalizati on alone is not enough to address the high dimensionality and high latency issues of the datas et, as it does not reduce the number of features or the variance in the data. D: Using a multiple correspondence analysis (MCA) m odel is not suitable for numeric variables, as it is a technique that reduces the dimensionality of t he dataset by transforming the original categorical variables into a smaller set of new variables, call ed factors, that capture most of the inertia and information in the data. MCA is similar to PCA, but it is designed for nominal or ordinal variables, n ot for continuous or interval variables.", + "references": "1: Principal Component Analysis - Amazon SageMaker 2: How to Use PCA for Data Visualization and Improv ed Performance in Machine Learning | by Pratik Shukla | Towards Data Science 3: Principal Component Analysis (PCA) for Feature S election and some of its Pitfalls | by Nagesh Singh Chauhan | Towards Data Science 4: How to Reduce Dimensionality with PCA and Train a Support Vector Machine in Python | by James Briggs | Towards Data Science 5: Dimensionality Reduction and Its Applications | by Aniruddha Bhandari | Towards Data Science 6: Principal Component Analysis (PCA) in Python | b y Susan Li | Towards Data Science 7: Feature Engineering for Machine Learning | by Di panjan (DJ) Sarkar | Towards Data Science 8: Feature Engineering \" How to Engineer Features a nd How to Get Good at It | by Parul Pandey | Towards Data Science : [Feature Scaling for Machine Learning: Understand ing the Difference Between Normalization vs. Standardization | by Benjamin Obi Tayo Ph.D. | Towa rds Data Science] : [Why, How and When to Scale your Features | by Ge orge Seif | Towards Data Science] : [Normalization vs Dimensionality Reduction | by S aurabh Annadate | Towards Data Science] : [Multiple Correspondence Analysis - Amazon SageMa ker] : [Multiple Correspondence Analysis (MCA) | by Raul Eulogio | Towards Data Science]" + }, + { + "question": "A company wants to forecast the daily price of newl y launched products based on 3 years of data for older product prices, sales, and rebates. The time- series data has irregular timestamps and is missing some values. Data scientist must build a dataset to replace the missing values. The data scientist needs a solution that resamptes the data daily and exports the data for further modeling. Which solution will meet these requirements with th e LEAST implementation effort?", + "options": [ + "A. Use Amazon EMR Serveriess with PySpark.", + "B. Use AWS Glue DataBrew.", + "C. Use Amazon SageMaker Studio Data Wrangler.", + "D. Use Amazon SageMaker Studio Notebook with Pandas." + ], + "correct": "C. Use Amazon SageMaker Studio Data Wrangler.", + "explanation": "Amazon SageMaker Studio Data Wrangler is a visual d ata preparation tool that enables users to clean and normalize data without writing any code. Using Data Wrangler, the data scientist can easily import the time-series data from various sources, s uch as Amazon S3, Amazon Athena, or Amazon Redshift. Data Wrangler can automatically generate data insights and quality reports, which can help identify and fix missing values, outliers, and anom alies in the data. Data Wrangler also provides over 250 built-in transformations, such as resampling, i nterpolation, aggregation, and filtering, which can be applied to the data with a point-and-click inter face. Data Wrangler can also export the prepared data to different destinations, such as Amazon S3, Amazon SageMaker Feature Store, or Amazon SageMaker Pipelines, for further modeling and analy sis. Data Wrangler is integrated with Amazon SageMaker Studio, a web-based IDE for machine learn ing, which makes it easy to access and use the tool. Data Wrangler is a serverless and fully manag ed service, which means the data scientist does not need to provision, configure, or manage any inf rastructure or clusters. Option A is incorrect because Amazon EMR Serverless is a serverless option for running big data analytics applications using open-source frameworks , such as Apache Spark. However, using Amazon EMR Serverless would require the data scientist to write PySpark code to perform the data preparation tasks, such as resampling, imputation, and aggregation. This would require more implementation effort than using Data Wrangler, whi ch provides a visual and code-free interface for data preparation. Option B is incorrect because AWS Glue DataBrew is another visual data preparation tool that can be used to clean and normalize data without writing co de. However, DataBrew does not support timeseries data as a data type, and does not provide built-in transformations for resampling, interpolation, or aggregation of time-series data. Therefore, using DataBrew would not meet the requirements of the use case. Option D is incorrect because using Amazon SageMake r Studio Notebook with Pandas would also require the data scientist to write Python code to perform the data preparation tasks. Pandas is a popular Python library for data analysis and manipu lation, which supports time-series data and provides various methods for resampling, interpolat ion, and aggregation. However, using Pandas would require more implementation effort than using Data Wrangler, which provides a visual and code-free interface for data preparation.", + "references": "1: Amazon SageMaker Data Wrangler documentation 2: Amazon EMR Serverless documentation 3: AWS Glue DataBrew documentation 4: Pandas documentation" + }, + { + "question": "A data scientist is building a forecasting model fo r a retail company by using the most recent 5 years of sales records that are stored in a data warehous e. The dataset contains sales records for each of the company's stores across five commercial regions The data scientist creates a working dataset with StorelD. Region. Date, and Sales Amount as col umns. The data scientist wants to analyze yearly average sales for each region. The scientist also w ants to compare how each region performed compared to average sales across all commercial reg ions. Which visualization will help the data scientist be tter understand the data trend?", + "options": [ + "A. Create an aggregated dataset by using the Pandas GroupBy function to get average sales for each", + "B. Create an aggregated dataset by using the Pandas GroupBy function to get average sales for each", + "C. Create an aggregated dataset by using the Pandas GroupBy function to get average sales for each", + "D. Create an aggregated dataset by using the Pandas GroupBy function to get average sales for each" + ], + "correct": "D. Create an aggregated dataset by using the Pandas GroupBy function to get average sales for each", + "explanation": "The best visualization for this task is to create a bar plot, faceted by year, of average sales for ea ch region and add a horizontal line in each facet to r epresent average sales. This way, the data scientis t can easily compare the yearly average sales for eac h region with the overall average sales and see the trends over time. The bar plot also allows the data scientist to see the relative performance of each region within each year and across years. The other options are less effective because they either do not show the yearly trends, do not show t he overall average sales, or do not group the data by region.", + "references": "pandas.DataFrame.groupby \" pandas 2.1.4 documentati on pandas.DataFrame.plot.bar \" pandas 2.1.4 documentat ion Matplotlib - Bar Plot - Online Tutorials Library" + }, + { + "question": "A company uses sensors on devices such as motor eng ines and factory machines to measure parameters, temperature and pressure. The company w ants to use the sensor data to predict equipment malfunctions and reduce services outages. The Machine learning (ML) specialist needs to gathe r the sensors data to train a model to predict device malfunctions The ML spoctafst must ensure th at the data does not contain outliers before training the ..el. What can the ML specialist meet these requirements with the LEAST operational overhead?", + "options": [ + "A. Load the data into an Amazon SagcMaker Studio not ebook. Calculate the first and third quartile", + "B. Use an Amazon SageMaker Data Wrangler bias report to find outliers in the dataset Use a Data", + "C. Use an Amazon SageMaker Data Wrangler anomaly det ection visualization to find outliers in the", + "D. Use Amazon Lookout for Equipment to find and remo ve outliers from the dataset." + ], + "correct": "C. Use an Amazon SageMaker Data Wrangler anomaly det ection visualization to find outliers in the", + "explanation": "Amazon SageMaker Data Wrangler is a tool that helps data scientists and ML developers to prepare data for ML. One of the features of Data Wrangler i s the anomaly detection visualization, which uses an unsupervised ML algorithm to identify outliers i n the dataset based on statistical properties. The ML specialist can use this feature to quickly explo re the sensor data and find any anomalous values that may affect the model performance. The ML speci alist can then add a transformation to a Data Wrangler data flow to remove the outliers from the dataset. The data flow can be exported as a script or a pipeline to automate the data preparation proc ess. This option requires the least operational overhead compared to the other options.", + "references": "Amazon SageMaker Data Wrangler - Amazon Web Service s (AWS) Anomaly Detection Visualization - Amazon SageMaker Transform Data - Amazon SageMaker" + }, + { + "question": "A data engineer needs to provide a team of data sci entists with the appropriate dataset to run machine learning training jobs. The data will be st ored in Amazon S3. The data engineer is obtaining the data from an Amazon Redshift database and is us ing join queries to extract a single tabular dataset. A portion of the schema is as follows: ...traction Timestamp (Timeslamp) ...JName(Varchar) ...JNo (Varchar) Th data engineer must provide the data so that any row with a CardNo value of NULL is removed. Also, the TransactionTimestamp column must be separ ated into a TransactionDate column and a isactionTime column Finally, the CardName column mu st be renamed to NameOnCard. The data will be extracted on a monthly basis and w ill be loaded into an S3 bucket. The solution must minimize the effort that is needed to set up infras tructure for the ingestion and transformation. The solution must be automated and must minimize the lo ad on the Amazon Redshift cluster Which solution meets these requirements?", + "options": [ + "A. Set up an Amazon EMR cluster Create an Apache Spa rk job to read the data from the Amazon", + "B. Set up an Amazon EC2 instance with a SQL client t ool, such as SQL Workbench/J. to query the data", + "C. Set up an AWS Glue job that has the Amazon Redshi ft cluster as the source and the S3 bucket as", + "D. Use Amazon Redshift Spectrum to run a query that writes the data directly to the S3 bucket." + ], + "correct": "C. Set up an AWS Glue job that has the Amazon Redshi ft cluster as the source and the S3 bucket as", + "explanation": "The best solution for this scenario is to set up an AWS Glue job that has the Amazon Redshift cluster as the source and the S3 bucket as the destination, and use the built-in transforms Filter, Map, and RenameField to perform the required transformations . This solution has the following advantages: It minimizes the effort that is needed to set up in frastructure for the ingestion and transformation, as AWS Glue is a fully managed service that provides a serverless Apache Spark environment, a graphical interface to define data sources and targ ets, and a code generation feature to create and edit scripts1. It automates the extraction and transformation proc ess, as AWS Glue can schedule the job to run monthly, and handle the connection, authentication, and configuration of the Amazon Redshift cluster and the S3 bucket2. It minimizes the load on the Amazon Redshift cluste r, as AWS Glue can read the data from the cluster in parallel and use a JDBC connection that supports SSL encryption3. It performs the required transformations, as AWS Gl ue can use the built-in transforms Filter, Map, and RenameField to remove the rows with NULL values , split the timestamp column into date and time columns, and rename the card name column, resp ectively4. The other solutions are not optimal or suitable, be cause they have the following drawbacks: A: Setting up an Amazon EMR cluster and creating an Apache Spark job to read the data from the Amazon Redshift cluster and transform the data is n ot the most efficient or convenient solution, as it requires more effort and resources to provision, co nfigure, and manage the EMR cluster, and to write and maintain the Spark code5. B: Setting up an Amazon EC2 instance with a SQL cli ent tool to query the data from the Amazon Redshift cluster directly and export the resulting dataset into a CSV file is not a scalable or reliab le solution, as it depends on the availability and per formance of the EC2 instance, and the manual execution and upload of the SQL queries and the CSV file6. D: Using Amazon Redshift Spectrum to run a query th at writes the data directly to the S3 bucket and creating an AWS Lambda function to run the query mo nthly is not a feasible solution, as Amazon Redshift Spectrum does not support writing data to external tables or S3 buckets, only reading data from them7.", + "references": "1: What Is AWS Glue? - AWS Glue 2: Populating the Data Catalog - AWS Glue 3: Best Practices When Using AWS Glue with Amazon R edshift - AWS Glue 4: Built-In Transforms - AWS Glue 5: What Is Amazon EMR? - Amazon EMR 6: Amazon EC2 - Amazon Web Services (AWS) 7: Using Amazon Redshift Spectrum to Query External Data - Amazon Redshift" + }, + { + "question": "A data scientist obtains a tabular dataset that con tains 150 correlated features with different ranges to build a regression model. The data scientist nee ds to achieve more efficient model training by implementing a solution that minimizes impact on th e model's performance. The data scientist decides to perform a principal component analysis ( PCA) preprocessing step to reduce the number of features to a smaller set of independent features b efore the data scientist uses the new features in the regression model. Which preprocessing step will meet these requiremen ts?", + "options": [ + "A. Use the Amazon SageMaker built-in algorithm for P CA on the dataset to transform the data", + "B. Load the data into Amazon SageMaker Data Wrangler . Scale the data with a Min Max Scaler", + "C. Reduce the dimensionality of the dataset by remov ing the features that have the highest", + "D. Reduce the dimensionality of the dataset by remov ing the features that have the lowest", + "A. Amazon" + ], + "correct": "B. Load the data into Amazon SageMaker Data Wrangler . Scale the data with a Min Max Scaler", + "explanation": "Principal component analysis (PCA) is a technique f or reducing the dimensionality of datasets, increasing interpretability but at the same time mi nimizing information loss. It does so by creating new uncorrelated variables that successively maximi ze variance. PCA is useful when dealing with datasets that have a large number of correlated fea tures. However, PCA is sensitive to the scale of the features, so it is important to standardize or normalize the data before applying PC", + "references": "" + }, + { + "question": "A financial services company wants to automate its loan approval process by building a machine learning (ML) model. Each loan data point contains credit history from a third-party data source and demographic information about the customer. Each lo an approval prediction must come with a report that contains an explanation for why the cus tomer was approved for a loan or was denied for a loan. The company will use Amazon SageMaker to bu ild the model. Which solution will meet these requirements with th e LEAST development effort?", + "options": [ + "A. Use SageMaker Model Debugger to automatically deb ug the predictions, generate the", + "B. Use AWS Lambda to provide feature importance and partial dependence plots. Use the plots to", + "C. Use SageMaker Clarify to generate the explanation report. Attach the report to the predicted", + "D. Use custom Amazon Cloud Watch metrics to generate the explanation report. Attach the report to" + ], + "correct": "C. Use SageMaker Clarify to generate the explanation report. Attach the report to the predicted", + "explanation": "The best solution for this scenario is to use SageM aker Clarify to generate the explanation report and attach it to the predicted results. SageMaker Clari fy provides tools to help explain how machine learning (ML) models make predictions using a model -agnostic feature attribution approach based on SHAP values. It can also detect and measure pote ntial bias in the data and the model. SageMaker Clarify can generate explanation reports during dat a preparation, model training, and model deployment. The reports include metrics, graphs, an d examples that help understand the model behavior and predictions. The reports can be attach ed to the predicted results using the SageMaker SDK or the SageMaker API. The other solutions are less optimal because they r equire more development effort and additional services. Using SageMaker Model Debugger would requ ire modifying the training script to save the model output tensors and writing custom rules to de bug and explain the predictions. Using AWS Lambda would require writing code to invoke the ML model, compute the feature importance and partial dependence plots, and generate and attach t he explanation report. Using custom Amazon CloudWatch metrics would require writing code to pu blish the metrics, create dashboards, and generate and attach the explanation report.", + "references": "Bias Detection and Model Explainability - Amazon Sa geMaker Clarify - AWS Amazon SageMaker Clarify Model Explainability Amazon SageMaker Clarify: Machine Learning Bias Det ection and Explainability GitHub - aws/amazon-sagemaker-clarify: Fairness Awa re Machine Learning" + }, + { + "question": "An obtain relator collects the following data on cu stomer orders: demographics, behaviors, location, shipment progress, and delivery time. A data scient ist joins all the collected datasets. The result is a single dataset that includes 980 variables. The data scientist must develop a machine learning (ML) model to identify groups of customers who are likely to respond to a marketing campaign. Which combination of algorithms should the data sci entist use to meet this requirement? (Select TWO.)", + "options": [ + "A. Latent Dirichlet Allocation (LDA)", + "B. K-means C. Se mantic feg mentation", + "D. Principal component analysis (PCA)" + ], + "correct": "", + "explanation": "The data scientist should use K-means and principal component analysis (PCA) to meet this requirement. K-means is a clustering algorithm that can group customers based on their similarity in the feature space. PCA is a dimensionality reductio n technique that can transform the original 980 variables into a smaller set of uncorrelated variab les that capture most of the variance in the data. This can help reduce the computational cost and noi se in the data, and improve the performance of the clustering algorithm.", + "references": "Clustering - Amazon SageMaker Dimensionality Reduction - Amazon SageMaker" + }, + { + "question": "A companys data scientist has trained a new machine learning model that performs better on test data than the companys existing model performs in t he production environment. The data scientist wants to replace the existing model that runs on an Amazon SageMaker endpoint in the production environment. However, the company is concerned that the new model might not work well on the production environment data. The data scientist needs to perform A/B testing in the production environment to evaluate whether the new model performs well on production environme nt data. Which combination of steps must the data scientist take to perform the A/B testing? (Choose two.)", + "options": [ + "A. Create a new endpoint configuration that includes a production variant for each of the two", + "B. Create a new endpoint configuration that includes two target variants that point to different", + "C. Deploy the new model to the existing endpoint.", + "D. Update the existing endpoint to activate the new model." + ], + "correct": "", + "explanation": "The combination of steps that the data scientist mu st take to perform the A/B testing are to create a new endpoint configuration that includes a producti on variant for each of the two models, and update the existing endpoint to use the new endpoin t configuration. This approach will allow the data scientist to deploy both models on the same en dpoint and split the inference traffic between them based on a specified distribution. Amazon SageMaker is a fully managed service that pr ovides developers and data scientists the ability to quickly build, train, and deploy machine learnin g models. Amazon SageMaker supports A/B testing on machine learning models by allowing the data scientist to run multiple production variants on an endpoint. A production variant is a version o f a model that is deployed on an endpoint. Each production variant has a name, a machine learning m odel, an instance type, an initial instance count, and an initial weight. The initial weight determine s the percentage of inference requests that the variant will handle. For example, if there are two variants with weights of 0.5 and 0.5, each variant will handle 50% of the requests. The data scientist can use production variants to test models that have been trained using different training datasets , algorithms, and machine learning frameworks; test how they perform on different instance types; or a combination of all of the above1. To perform A/B testing on machine learning models, the data scientist needs to create a new endpoint configuration that includes a production v ariant for each of the two models. An endpoint configuration is a collection of settings that defi ne the properties of an endpoint, such as the name, the production variants, and the data capture confi guration. The data scientist can use the Amazon SageMaker console, the AWS CLI, or the AWS SDKs to create a new endpoint configuration. The data scientist needs to specify the name, model name, in stance type, initial instance count, and initial variant weight for each production variant in the e ndpoint configuration2. After creating the new endpoint configuration, the data scientist needs to update the existing endpoint to use the new endpoint configuration. Upd ating an endpoint is the process of deploying a new endpoint configuration to an existing endpoint. Updating an endpoint does not affect the availability or scalability of the endpoint, as Ama zon SageMaker creates a new endpoint instance with the new configuration and switches the DNS rec ord to point to the new instance when it is ready. The data scientist can use the Amazon SageMa ker console, the AWS CLI, or the AWS SDKs to update an endpoint. The data scientist needs to spe cify the name of the endpoint and the name of the new endpoint configuration to update the endpoi nt3. The other options are either incorrect or unnecessa ry. Creating a new endpoint configuration that includes two target variants that point to differen t endpoints is not possible, as target variants are only used to invoke a specific variant on an endpoi nt, not to define an endpoint configuration. Deploying the new model to the existing endpoint wo uld replace the existing model, not run it sideby- side with the new model. Updating the existing endp oint to activate the new model is not a valid operation, as there is no activation parameter for an endpoint.", + "references": "1: A/B Testing ML models in production using Amazon SageMaker | AWS Machine Learning Blog 2: Create an Endpoint Configuration - Amazon SageMa ker 3: Update an Endpoint - Amazon SageMaker" + }, + { + "question": "An online store is predicting future book sales by using a linear regression model that is based on past sales data. The data includes duration, a nume rical feature that represents the number of days that a book has been listed in the online store. A data scientist performs an exploratory data analysi s and discovers that the relationship between book sa les and duration is skewed and non-linear. Which data transformation step should the data scie ntist take to improve the predictions of the model?", + "options": [ + "A. One-hot encoding", + "B. Cartesian product transformation", + "C. Quantile binning", + "D. Normalization", + "A. Use an object detection algorithm to identify a v isitors hair in video frames. Pass the identified", + "B. Use an object detection algorithm to identify a v isitors hair in video frames. Pass the identified", + "C. Use a semantic segmentation algorithm to identify a visitors hair in video frames. Pass the", + "D. Use a semantic segmentation algorithm to identify a visitors hair in video frames. Pass the" + ], + "correct": "C. Use a semantic segmentation algorithm to identify a visitors hair in video frames. Pass the", + "explanation": "The solution that will meet the requirements with t he least amount of effort is to use a semantic segmentation algorithm to identify a visitors hair in video frames, and pass the identified hair to an ResNet-50 algorithm to determine hair style and hai r color. This solution can leverage the existing Amazon SageMaker algorithms and frameworks to perfo rm the tasks of hair segmentation and classification. Semantic segmentation is a computer vision techniqu e that assigns a class label to every pixel in an image, such that pixels with the same label share c ertain characteristics. Semantic segmentation can be used to identify and isolate different objects o r regions in an image, such as a visitors hair in a video frame. Amazon SageMaker provides a built-in s emantic segmentation algorithm that can train and deploy models for semantic segmentation tasks. The algorithm supports three state-of-the-art network architectures: Fully Convolutional Network (FCN), Pyramid Scene Parsing Network (PSP), and DeepLab v3. The algorithm can also use pre-trai ned or randomly initialized ResNet-50 or ResNet- 101 as the backbone network. The algorithm can be t rained using P2/P3 type Amazon EC2 instances in single machine configurations1. ResNet-50 is a convolutional neural network that is 50 layers deep and can classify images into 1000 object categories. ResNet-50 is trained on more tha n a million images from the ImageNet database and can achieve high accuracy on various image reco gnition tasks. ResNet-50 can be used to determine hair style and hair color from the segmen ted hair regions in the video frames. Amazon SageMaker provides a built-in image classification algorithm that can use ResNet-50 as the network architecture. The algorithm can also perform transf er learning by fine-tuning the pre-trained ResNet- 50 model with new data. The algorithm can be traine d using P2/P3 type Amazon EC2 instances in single or multiple machine configurations2. The other options are either less effective or more complex to implement. Using an object detection algorithm to identify a visitors hair in video fram es would not segment the hair at the pixel level, b ut only draw bounding boxes around the hair regions. T his could result in inaccurate or incomplete hair segmentation, especially if the hair is occluded or has irregular shapes. Using an XGBoost algorithm to determine hair style and hair color would requir e transforming the segmented hair images into numerical features, which could lose some informati on or introduce noise. XGBoost is also not designed for image classification tasks, and may no t achieve high accuracy or performance.", + "references": "1: Semantic Segmentation Algorithm - Amazon SageMak er 2: Image Classification Algorithm - Amazon SageMake r" + }, + { + "question": "A company wants to predict stock market price trend s. The company stores stock market data each business day in Amazon S3 in Apache Parquet format. The company stores 20 GB of data each day for each stock code. A data engineer must use Apache Spark to perform ba tch preprocessing data transformations quickly so the company can complete prediction jobs before the stock market opens the next day. The company plans to track more stock market codes and needs a way to scale the preprocessing data transformations. Which AWS service or feature will meet these requir ements with the LEAST development effort over time?", + "options": [ + "A. AWS Glue jobs", + "B. Amazon EMR cluster", + "C. Amazon Athena", + "D. AWS Lambda" + ], + "correct": "A. AWS Glue jobs", + "explanation": "AWS Glue jobs is the AWS service or feature that wi ll meet the requirements with the least development effort over time. AWS Glue jobs is a fu lly managed service that enables data engineers to run Apache Spark applications on a serverless Sp ark environment. AWS Glue jobs can perform batch preprocessing data transformations on large d atasets stored in Amazon S3, such as converting data formats, filtering data, joining data, and agg regating data. AWS Glue jobs can also scale the Spark environment automatically based on the data v olume and processing needs, without requiring any infrastructure provisioning or management. AWS Glue jobs can reduce the development effort and time by providing a graphical interface to crea te and monitor Spark applications, as well as a code generation feature that can generate Scala or Python code based on the data sources and targets. AWS Glue jobs can also integrate with othe r AWS services, such as Amazon Athena, Amazon EMR, and Amazon SageMaker, to enable further data a nalysis and machine learning tasks1. The other options are either more complex or less s calable than AWS Glue jobs. Amazon EMR cluster is a managed service that enables data engineers to run Apache Spark applications on a cluster of Amazon EC2 instances. However, Amazon EMR cluster r equires more development effort and time than AWS Glue jobs, as it involves setting up, conf iguring, and managing the cluster, as well as writing and deploying the Spark code. Amazon EMR cl uster also does not scale automatically, but requires manual or scheduled resizing of the cluste r based on the data volume and processing needs2. Amazon Athena is a serverless interactive q uery service that enables data engineers to analyze data stored in Amazon S3 using standard SQL . However, Amazon Athena is not suitable for performing complex data transformations, such as jo ining data from multiple sources, aggregating data, or applying custom logic. Amazon Athena is al so not designed for running Spark applications, but only supports SQL queries3. AWS Lambda is a ser verless compute service that enables data engineers to run code without provisioning or manag ing servers. However, AWS Lambda is not optimized for running Spark applications, as it has limitations on the execution time, memory size, and concurrency of the functions. AWS Lambda is als o not integrated with Amazon S3, and requires additional steps to read and write data from S3 buc kets.", + "references": "1: AWS Glue - Fully Managed ETL Service - Amazon We b Services 2: Amazon EMR - Amazon Web Services 3: Amazon Athena \" Interactive SQL Queries for Data in Amazon S3 [4]: AWS Lambda \" Serverless Compute - Amazon Web S ervices" + }, + { + "question": "A company wants to enhance audits for its machine l earning (ML) systems. The auditing system must be able to perform metadata analysis on the feature s that the ML models use. The audit solution must generate a report that analyzes the metadata. The solution also must be able to set the data sensitivity and authorship of features. Which solution will meet these requirements with th e LEAST development effort?", + "options": [ + "A. Use Amazon SageMaker Feature Store to select the features. Create a data flow to perform", + "B. Use Amazon SageMaker Feature Store to set feature groups for the current features that the ML", + "C. Use Amazon SageMaker Features Store to apply cust om algorithms to analyze the feature-level", + "D. Use Amazon SageMaker Feature Store to set feature groups for the current features that the ML" + ], + "correct": "D. Use Amazon SageMaker Feature Store to set feature groups for the current features that the ML", + "explanation": "The solution that will meet the requirements with t he least development effort is to use Amazon SageMaker Feature Store to set feature groups for t he current features that the ML models use, assign the required metadata for each feature, and use Amazon QuickSight to analyze the metadata. This solution can leverage the existing AWS service s and features to perform feature-level metadata analysis and reporting. Amazon SageMaker Feature Store is a fully managed, purpose-built repository to store, update, search, and share machine learning (ML) features. T he service provides feature management capabilities such as enabling easy feature reuse, l ow latency serving, time travel, and ensuring consistency between features used in training and i nference workflows. A feature group is a logical grouping of ML features whose organization and stru cture is defined by a feature group schema. A feature group schema consists of a list of feature definitions, each of which specifies the name, type , and metadata of a feature. The metadata can include information such as data sensitivity, authorship, description, and parameters. The metada ta can help make features discoverable, understandable, and traceable. Amazon SageMaker Fea ture Store allows users to set feature groups for the current features that the ML models use, an d assign the required metadata for each feature using the AWS SDK for Python (Boto3), AWS Command L ine Interface (AWS CLI), or Amazon SageMaker Studio1. Amazon QuickSight is a fully managed, serverless bu siness intelligence service that makes it easy to create and publish interactive dashboards that incl ude ML insights. Amazon QuickSight can connect to various data sources, such as Amazon S3, Amazon Athena, Amazon Redshift, and Amazon SageMaker Feature Store, and analyze the data using standard SQL or built-in ML-powered analytics. Amazon QuickSight can also create rich visualizatio ns and reports that can be accessed from any device, and securely shared with anyone inside or o utside an organization. Amazon QuickSight can be used to analyze the metadata of the features sto red in Amazon SageMaker Feature Store, and generate a report that summarizes the metadata anal ysis2. The other options are either more complex or less e ffective than the proposed solution. Using Amazon SageMaker Data Wrangler to select the featur es and create a data flow to perform featurelevel metadata analysis would require additional steps an d resources, and may not capture all the metadata attributes that the company requires. Crea ting an Amazon DynamoDB table to store feature-level metadata would introduce redundancy a nd inconsistency, as the metadata is already stored in Amazon SageMaker Feature Store. Using Sag eMaker Studio to analyze the metadata would not generate a report that can be easily shared and accessed by the company.", + "references": "1: Amazon SageMaker Feature Store \" Amazon Web Serv ices 2: Amazon QuickSight \" Business Intelligence Servic e - Amazon Web Services" + }, + { + "question": "A machine learning (ML) engineer has created a feat ure repository in Amazon SageMaker Feature Store for the company. The company has AWS accounts for development, integration, and production. The company hosts a feature store in th e development account. The company uses Amazon S3 buckets to store feature values offline. The company wants to share features and to allow the integration account and the production account to reuse the features that are in the feature repos itory. Which combination of steps will meet these requirem ents? (Select TWO.)", + "options": [ + "A. Create an IAM role in the development account tha t the integration account and production", + "B. Share the feature repository that is associated t he S3 buckets from the development account to", + "C. Use AWS Security Token Service (AWS STS) from the integration account and the production account to", + "D. Set up S3 replication between the development S3 buckets and the integration and production S3 bucke ts." + ], + "correct": "", + "explanation": "The combination of steps that will meet the require ments are to create an IAM role in the development account that the integration account an d production account can assume, attach IAM policies to the role that allow access to the featu re repository and the S3 buckets, and share the feature repository that is associated with the S3 b uckets from the development account to the integration account and the production account by u sing AWS Resource Access Manager (AWS RAM). This approach will enable cross-account access and sharing of the features stored in Amazon SageMaker Feature Store and Amazon S3. Amazon SageMaker Feature Store is a fully managed, purpose-built repository to store, update, search, and share curated data used in training and prediction workflows. The service provides feature management capabilities such as enabling ea sy feature reuse, low latency serving, time travel, and ensuring consistency between features u sed in training and inference workflows. A feature group is a logical grouping of ML features whose organization and structure is defined by a feature group schema. A feature group schema consis ts of a list of feature definitions, each of which specifies the name, type, and metadata of a feature . Amazon SageMaker Feature Store stores the features in both an online store and an offline sto re. The online store is a low-latency, highthroughp ut store that is optimized for real-time inference. Th e offline store is a historical store that is backed by an Amazon S3 bucket and is optimized for batch processing and model training1. AWS Identity and Access Management (IAM) is a web s ervice that helps you securely control access to AWS resources for your users. You use IAM to con trol who can use your AWS resources (authentication) and what resources they can use an d in what ways (authorization). An IAM role is an IAM identity that you can create in your account th at has specific permissions. You can use an IAM role to delegate access to users, applications, or services that dont normally have access to your AWS resources. For example, you can create an IAM r ole in your development account that allows the integration account and the production account to assume the role and access the resources in the development account. You can attach IAM policie s to the role that specify the permissions for the feature repository and the S3 buckets. You can also use IAM conditions to restrict the access based on the source account, IP address, or other factors 2. AWS Resource Access Manager (AWS RAM) is a service that enables you to easily and securely share AWS resources with any AWS account or within your A WS Organization. You can share AWS resources that you own with other accounts using resource sha res. A resource share is an entity that defines the resources that you want to share, and the princ ipals that you want to share with. For example, you can share the feature repository that is associ ated with the S3 buckets from the development account to the integration account and the producti on account by creating a resource share in AWS RAM. You can specify the feature group ARN and the S3 bucket ARN as the resources, and the integration account ID and the production account I D as the principals. You can also use IAM policies to further control the access to the shared resourc es3. The other options are either incorrect or unnecessa ry. Using AWS Security Token Service (AWS STS) from the integration account and the production acc ount to retrieve credentials for the development account is not required, as the IAM role in the dev elopment account can provide temporary security credentials for the cross-account access. Setting u p S3 replication between the development S3 buckets and the integration and production S3 bucke ts would introduce redundancy and inconsistency, as the S3 buckets are already shared through AWS RAM. Creating an AWS PrivateLink endpoint in the development account for SageMaker i s not relevant, as it is used to securely connect to SageMaker services from a VPC, not from another account.", + "references": "1: Amazon SageMaker Feature Store \" Amazon Web Serv ices 2: What Is IAM? - AWS Identity and Access Managemen t 3: What Is AWS Resource Access Manager? - AWS Resou rce Access Manager 3: What Is AWS Resource Access Manager? - AWS Resou rce Access Manager" + } +] \ No newline at end of file