| Description | |
| Data Science London is hosting a meetup on Scikit-learn. This competition is a practice ground for trying, sharing, and creating examples of sklearn's classification abilities. If this turns into something useful, we can follow it up with regression or more complex classification problems. We encourage participants to post code via the "Tutorials" link on the left. Don't worry about accuracy or whether your code is perfect. The aim here is to explore sklearn by using it. You do not need to use sklearn to enter the competition. If you're new, we hope you'll use this opportunity to practice a new tool. If you're an expert, we hope you'll share the knowledge and document interesting ways to approach this problem. | |
| Scikit-learn (sklearn) is an established, open-source machine learning library, written in Python with the help of NumPy, SciPy, and Cython. Scikit-learn is very user-friendly, has a consistent API, and provides extensive documentation. Its implementation is high quality due to strict coding standards and high test coverage. Behind sklearn is a very active community, which is steadily improving the library. | |
| Meetup Information | |
| Thursday, March 7, 2013, 6:30 PM UTC | |
| http://www.meetup.com/Data-Science-London/events/105840372/ | |
| “Learning in Python with scikit-learn" by Andreas Mueller | |
| This talk will give an overview of the library and introduce general machine learning concepts such as supervised and unsupervised learning, feature extraction, cross-validation for model evaluation, and hyperparameter selection. We will also touch on some more advanced yet practically useful concepts such as feature hashing and ensemble learning. | |
| Andreas is a PhD student in machine learning and computer vision at Bonn University in Germany. He is one of the core developers and the maintainer of scikit-learn and the author of the blog peekaboo-vision. His interests include principles and applications of machine learning and open science. | |
| "Parallel and large scale learning with scikit-learn" by Olivier Grisel | |
| This talk will introduce practical tools and concepts to better leverage multicore machines and small clusters to perform interactive yet scalable predictive modeling with scikit-learn and IPython.parallel. In particular, we will introduce: | |
| - A short introduction to the parallel features of IPython from the notebook interface | |
| - How to perform scalable text feature extraction with the Hashing Trick | |
| - How to parallelize or distribute model evaluation (cross-validation) and hyperparameters tuning | |
| - How to optimize memory usage with memory mapping | |
| - How to approximate kernel Support Vector Machines for large scale datasets | |
| - A short introduction to Ensembles with model averaging and Random Forests | |
| Olivier is an R&D Software Engineer working in Java by day and a Python machine learning hacker by night. He is interested in applications to Natural Language Processing, Computer Vision, and predictive modeling. | |
| Evaluation | |
| This is a binary classification task. You are evaluated on classification accuracy (the percentage of labels you predict correctly). The training set has 1000 samples and the testing set has 9000. Your prediction should be a 9000 x 1 vector of ones or zeros. You also need an Id column (1 to 9000) and should include a header. The format looks like this: | |
| Id,Solution | |
| 1,0 | |
| 2,1 | |
| 3,1 | |
| ... | |
| 9000,0 | |
| Dataset Description | |
| This is a synthetic dataset of 40 features, representing objects from two classes (labeled as 0 or 1). The training set has 1000 samples and the testing set has 9000. |