SWE-Dev/SWE-Dev-3 · Datasets at Hugging Face

repo stringlengths 7 59	instance_id stringlengths 11 63	base_commit stringlengths 40 40	patch stringlengths 167 798k	problem_statement stringlengths 20 65.2k	created_at timestamp[ns]date 2015-08-30 10:31:05 2024-12-13 16:08:19	FAIL_TO_PASS sequencelengths 0 0	PASS_TO_PASS sequencelengths 0 0
augerai/a2ml	augerai__a2ml-611	30601dc95093e45ebafb500fa70fcdafe65edd24	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index e71bfca8..5baef1cb 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.79' +__version__ = '1.0.80' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 03454e01..b7002dab 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -279,7 +279,7 @@ def deploy(self, model_id, locally=False, review=False, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, threshold=None, score=False, score_true_data=None, - output=None, no_features_in_result = None, locally=False, provider=None): + output=None, no_features_in_result = None, locally=False, provider=None, predict_labels=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -299,6 +299,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. + predict_labels (dict, bool): Run ActiveLearn to select data for labelling Returns: if filename is not None. :: @@ -363,7 +364,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at """ return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result, - score, score_true_data ) + score, score_true_data, predict_labels ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 7f810fd8..770c9376 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -68,7 +68,7 @@ def deploy(self, model_id, locally=False, review=False, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, threshold=None, score=False, score_true_data=None, - output=None, no_features_in_result=None, locally=False, provider=None): + output=None, no_features_in_result=None, locally=False, provider=None, predict_labels=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -88,6 +88,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. + predict_labels (dict, bool): Run ActiveLearn to select data for labelling Returns: if filename is not None. :: @@ -152,7 +153,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at """ return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result, - score, score_true_data ) + score, score_true_data, predict_labels ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index f367c7a6..af40e9f8 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -25,10 +25,11 @@ def deploy(self, model_id, locally=False, review=False, name=None, algorithm=Non return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path, metadata) def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, - predicted_at=None, output=None, no_features_in_result=None, score=False, score_true_data=None): + predicted_at=None, output=None, no_features_in_result=None, score=False, + score_true_data=None, predict_labels=None): return AugerModel(self.ctx).predict( model_id, filename, threshold, locally, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, experiment_params=None, locally=False): diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index dc056474..aacf1a0e 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -95,7 +95,7 @@ def has_endpoint(self, props=None): return is_endpoint def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, - no_features_in_result=None, score=False, score_true_data=None): + no_features_in_result=None, score=False, score_true_data=None, predict_labels=None): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') @@ -120,7 +120,7 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at prediction_properties = \ prediction_api.create(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at, no_features_in_result=no_features_in_result, - score=score, score_true_data=score_true_data) + score=score, score_true_data=score_true_data, predict_labels=predict_labels) return prediction_properties.get('result') diff --git a/a2ml/api/auger/impl/cloud/prediction.py b/a2ml/api/auger/impl/cloud/prediction.py index 81f50387..da3dd44b 100644 --- a/a2ml/api/auger/impl/cloud/prediction.py +++ b/a2ml/api/auger/impl/cloud/prediction.py @@ -15,7 +15,7 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self._set_api_request_path("AugerEndpointPredictionApi") def create(self, records, features, threshold=None, file_url=None, predicted_at=None, - no_features_in_result=None, score=False, score_true_data=None): + no_features_in_result=None, score=False, score_true_data=None, predict_labels=None): params = { 'records': records, 'features': features, @@ -43,4 +43,7 @@ def create(self, records, features, threshold=None, file_url=None, predicted_at= if score_true_data: params['score_true_data'] = score_true_data + if predict_labels: + params['predict_labels'] = predict_labels + return self._call_create(params, ['requested', 'running']) diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 1af255e6..f2797bb4 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -31,9 +31,9 @@ def undeploy(self, model_id, locally=False): return ModelUndeploy(self.ctx, self.project).execute(model_id, locally) def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, - output=None, no_features_in_result=None, score=False, score_true_data=None): + output=None, no_features_in_result=None, score=False, score_true_data=None, predict_labels=None): return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, - predicted_at, output, no_features_in_result, score, score_true_data) + predicted_at, output, no_features_in_result, score, score_true_data, predict_labels) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, experiment_params=None, locally=False): diff --git a/a2ml/api/auger/impl/mparts/predict.py b/a2ml/api/auger/impl/mparts/predict.py index d523f6ca..8743c065 100644 --- a/a2ml/api/auger/impl/mparts/predict.py +++ b/a2ml/api/auger/impl/mparts/predict.py @@ -24,7 +24,7 @@ def __init__(self, ctx): def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None, - score=False, score_true_data=None): + score=False, score_true_data=None, predict_labels=None): if filename is not None and isinstance(filename, str) and \ not (filename.startswith("http:") or filename.startswith("https:")) and \ not fsclient.is_s3_path(filename): @@ -34,13 +34,13 @@ def execute(self, filename, model_id, threshold=None, locally=False, data=None, if locally: if locally == "docker": predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) else: predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) else: predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) return predicted @@ -104,7 +104,7 @@ def _check_model_project(self, pipeline_api): self.ctx.config.get('name'), model_project_name)) def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, - output, no_features_in_result, score, score_true_data): + output, no_features_in_result, score, score_true_data, predict_labels): records, features, file_url, is_pandas_df = self._process_input(filename, data, columns) temp_file = None ds_result = None @@ -114,7 +114,7 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic pipeline_api = AugerPipelineApi(self.ctx, None, model_id) predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at, no_features_in_result=no_features_in_result, - score=score, score_true_data=score_true_data) + score=score, score_true_data=score_true_data, predict_labels=predict_labels) try: ds_result = DataFrame.create_dataframe(predictions.get('signed_prediction_url'), @@ -145,7 +145,7 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic fsclient.remove_file(temp_file) def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, - output, no_features_in_result, score, score_true_data): + output, no_features_in_result, score, score_true_data, predict_labels): from auger_ml.model_exporter import ModelExporter is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) @@ -161,12 +161,17 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre if score and score_true_data is None: options = fsclient.read_json_file(os.path.join(model_path, "options.json")) - ds = DataFrame.create_dataframe(filename_arg, data, [options['targetFeature']]) + ds = DataFrame.create_dataframe(filename_arg, data)#, [options['targetFeature']]) score_true_data = ds.df - - res, options = ModelExporter({}).predict_by_model_to_ds(model_path, - path_to_predict=filename_arg, records=data, features=columns, - threshold=threshold, no_features_in_result=no_features_in_result) + + if predict_labels: + res, options = ModelExporter({}).predict_labels_by_model_to_ds(model_path, + path_to_predict=filename_arg, records=data, features=columns, + threshold=threshold, no_features_in_result=no_features_in_result, predict_labels=predict_labels) + else: + res, options = ModelExporter({}).predict_by_model_to_ds(model_path, + path_to_predict=filename_arg, records=data, features=columns, + threshold=threshold, no_features_in_result=no_features_in_result) ds_result = DataFrame({'data_path': None}) ds_result.df = res.df @@ -192,7 +197,7 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre # no_features_in_result=no_features_in_result) #, output=output) def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, - output, no_features_in_result, score, score_true_data): + output, no_features_in_result, score, score_true_data, predict_labels): model_deploy = ModelDeploy(self.ctx, None) is_model_loaded, model_path = model_deploy.verify_local_model(model_id, add_model_folder=False) if not is_model_loaded: @@ -205,7 +210,7 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co filename = os.path.join(self.ctx.config.get_path(), '.augerml', 'predict_data.csv') ds.saveToCsvFile(filename, compression=None) - predicted = self._docker_run_predict(filename, threshold, model_path, score, score_true_data) + predicted = self._docker_run_predict(filename, threshold, model_path, score, score_true_data, predict_labels) if not filename_arg: ds_result = DataFrame.create_dataframe(predicted) @@ -223,7 +228,7 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co return predicted - def _docker_run_predict(self, filename, threshold, model_path, score, score_true_data): + def _docker_run_predict(self, filename, threshold, model_path, score, score_true_data, predict_labels): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) docker_tag = cluster_settings.get('kubernetes_stack') predict_file = os.path.basename(filename) diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 326a219f..5f061718 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -23,15 +23,15 @@ def deploy(self, project, model_id, locally, review, name, algorithm, score, dat @authenticated #@with_project(autocreate=False) def predict(self, filename, model_id, threshold, locally, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data): + no_features_in_result, score, score_true_data, predict_labels): if locally: self.deploy(model_id, locally, review=False, name=None, algorithm=None, score=None, data_path=None) predicted = Model(self.ctx, project=None).predict( filename, model_id, threshold, locally, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) - if filename: + if output: self.ctx.log('Predictions stored in %s' % predicted) if isinstance(predicted, dict) and 'predicted' in predicted: diff --git a/a2ml/api/utils/formatter.py b/a2ml/api/utils/formatter.py index e7a118bb..d5f37508 100644 --- a/a2ml/api/utils/formatter.py +++ b/a2ml/api/utils/formatter.py @@ -12,7 +12,8 @@ def print_table(log, table_list, headers=None, hor_lines=True): col_list = list(table_list[0].keys() if table_list else []) row_list = [col_list] # 1st row = header for item in table_list: - row_list.append([str(item.get(col) or '') for col in col_list]) + row_list.append([str(item.get(col)) if item.get(col) is not None else '' for col in col_list]) + # maximun size of the col for each element col_size = [max(map(len, col)) for col in zip(*row_list)] # insert seperating line before every line, and extra one for ending. diff --git a/setup.py b/setup.py index 646d8b7c..3998fdeb 100644 --- a/setup.py +++ b/setup.py @@ -84,13 +84,13 @@ def run(self): 'google-cloud-automl' ], 'predict': [ - 'auger.ai.predict[all]==1.0.104' + 'auger.ai.predict[all]==1.0.106' ], 'predict_no_cat_lgbm': [ - 'auger.ai.predict[no_cat_lgbm]==1.0.104' + 'auger.ai.predict[no_cat_lgbm]==1.0.106' ], 'predict_no_lgbm': [ - 'auger.ai.predict[no_cat_lgbm]==1.0.104', + 'auger.ai.predict[no_cat_lgbm]==1.0.106', 'catboost' ] }	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2022-09-29T10:02:12	[]	[]
augerai/a2ml	augerai__a2ml-610	0198cb1d42783000318a5e4579018a8be362743f	diff --git a/.circleci/config.yml b/.circleci/config.yml deleted file mode 100644 index 8b70a2d2..00000000 --- a/.circleci/config.yml +++ /dev/null @@ -1,254 +0,0 @@ -version: 2.1 - -orbs: - docker: circleci/[email protected] - aws-eks: circleci/[email protected] - role: airswap/[email protected] - aws-cli: circleci/[email protected] - kubernetes: circleci/[email protected] - helm: circleci/[email protected] - -jobs: - build-docs: - docker: - - image: circleci/python:3.7-stretch - steps: - - checkout - - run: sudo chown -R circleci:circleci /usr/local/bin - - run: sudo chown -R circleci:circleci /usr/local/lib/python3.7/site-packages - - restore_cache: - keys: - - a2ml-python-doc-deps-v3-{{ arch }}-3.7-{{ .Branch }}-{{ checksum "setup.py" }}-{{ checksum "docs/requirements.txt" }} - - a2ml-python-doc-deps-v3-{{ arch }}-3.7-{{ .Branch }} - - a2ml-python-doc-deps-v3-{{ arch }}-3.7 - - run: - name: Install dependencies - command: \| - virtualenv venv - source venv/bin/activate - make develop-docs - - save_cache: - key: a2ml-python-doc-deps-v3-{{ arch }}-3.7-{{ .Branch }}-{{ checksum "setup.py" }}-{{ checksum "docs/requirements.txt" }} - paths: - - "venv" - - "/home/circleci/.cache/pip" - - run: - name: Build docs - command: \| - source venv/bin/activate - cd docs/ - make html - - persist_to_workspace: - root: docs/build - paths: html - publish-docs: - docker: - - image: node:10.15.0 - steps: - - checkout - - attach_workspace: - at: docs/build - - add_ssh_keys: - fingerprints: "44:aa:23:95:60:12:6b:b5:8d:b2:e5:05:24:1f:94:cf" - - run: - name: Deploy docs to gh-pages branch - command: \| - git config user.email "[email protected]" - git config user.name "augerbot" - npm install -g --silent [email protected] - gh-pages --dotfiles --message "[skip ci] Updates" --dist docs/build/html - build-and-test: - docker: - - image: cimg/base:stable - steps: - - checkout - - setup_remote_docker - - run: - name: Install Docker Compose - environment: - COMPOSE_VERSION: '1.29.2' - command: \| - curl -L "https://github.com/docker/compose/releases/download/${COMPOSE_VERSION}/docker-compose-$(uname -s)-$(uname -m)" -o ~/docker-compose - chmod +x ~/docker-compose - sudo mv ~/docker-compose /usr/local/bin/docker-compose - - run: make config docker-test - - run: make docker-save - - persist_to_workspace: - root: . - paths: - - ./image.tar.gz - publish-docker: - executor: docker/machine - parameters: - docker-tag: - description: Tag to publish - type: string - default: latest - steps: - - checkout - - attach_workspace: - at: ./ - - docker/check - - run: make docker-load - - run: DOCKER_TAG=<< parameters.docker-tag >> make docker-tag - - docker/push: - image: augerai/a2ml - tag: << parameters.docker-tag >> - - publish-pip: - docker: - - image: circleci/python:3.7-stretch - steps: - - checkout - - run: - command: \| - echo -e "[pypi]" >> ~/.pypirc - echo -e "username = $PYPI_USERNAME" >> ~/.pypirc - echo -e "password = $PYPI_PASSWORD" >> ~/.pypirc - - run: make build - - run: make release - - deploy-to-k8s: - executor: aws-eks/python3 - parameters: - cluster-name: - description: \| - Name of the EKS cluster - type: string - default: ${STAGING_CLUSTER_NAME} - aws-region: - description: \| - AWS region - type: string - default: ${AWS_DEFAULT_REGION} - namespace: - description: \| - a2ml namespace - type: string - default: a2ml - release-name: - description: \| - a2ml helm release-name - type: string - default: a2ml - reuse-values: - description: \| - Reuse last release's values and merge in any overrides - type: boolean - default: true - account-id: - description: \| - AWS account containing the cluster - type: string - default: ${ACCOUNT_ID} - role-name: - description: \| - AWS role to assume for deploying to eks - type: string - default: ${ROLE_NAME} - values-to-override: - description: \| - Values will be used as helm install --set "key1=value1,key2=value2" - type: string - default: "" - steps: - - aws-cli/setup - - role/assume-role: - account-id: << parameters.account-id >> - role-name: << parameters.role-name >> - - aws-eks/update-kubeconfig-with-authenticator: - cluster-name: << parameters.cluster-name >> - aws-region: << parameters.aws-region >> - install-kubectl: true - - helm/install-helm-client: - version: v3.4.2 - - run: - command: \| - helm repo add augerai https://augerai.github.io/charts - helm repo update - helm repo list - name: Add augerai repo - - helm/upgrade-helm-chart: - helm-version: v3.2.4 - chart: augerai/a2ml - namespace: << parameters.namespace >> - release-name: << parameters.release-name >> - reuse-values: << parameters.reuse-values >> - values-to-override: << parameters.values-to-override >> - -workflows: - build-test-publish: - jobs: - - build-and-test: - filters: - tags: - only: /^v./ - - build-docs - - publish-docs: - requires: - - build-and-test - - build-docs - filters: - branches: - only: - - master - - publish-docker: - context: docker-hub - name: docker-publish-tag - docker-tag: $CIRCLE_TAG - requires: - - build-and-test - filters: - tags: - only: /^v./ - branches: - ignore: /./ - - publish-docker: - context: docker-hub - name: docker-publish-master - requires: - - build-and-test - filters: - branches: - only: - - master - tags: - ignore: /^v./ - - deploy-to-k8s: - name: deploy-to-k8s-master - cluster-name: ${STAGING_CLUSTER_NAME} - values-to-override: "image.tag=latest" - release-name: a2ml - namespace: a2ml - context: eks - requires: - - docker-publish-master - filters: - branches: - only: - - master - tags: - ignore: /^v./ - - deploy-to-k8s: - name: deploy-to-k8s - namespace: a2ml - release-name: a2ml - values-to-override: "image.tag=$CIRCLE_TAG" - context: eks - requires: - - docker-publish-tag - account-id: ${PROD_ACCOUNT_ID} - cluster-name: ${PROD_CLUSTER_NAME} - role-name: ${PROD_ROLE_NAME} - filters: - tags: - only: /^v./ - branches: - ignore: /./ - - publish-pip: - filters: - tags: - only: /^v./ - branches: - ignore: /./ - context: pypi diff --git a/.github/workflows/docs.yml b/.github/workflows/docs.yml new file mode 100644 index 00000000..f87c87d2 --- /dev/null +++ b/.github/workflows/docs.yml @@ -0,0 +1,32 @@ +name: docs + +on: + push: + tags: + - v + +jobs: + build_publish_docs: + runs-on: ubuntu-latest + + steps: + - uses: actions/checkout@v3 + - name: Set up Python + uses: actions/setup-python@v3 + with: + python-version: 3.7 + + - name: Build docs + run: \| + make develop-docs + cd docs/ + make html + + - name: Deploy + uses: peaceiris/actions-gh-pages@v3 + with: + github_token: ${{ secrets.GITHUB_TOKEN }} + publish_dir: docs/build/html + + + \ No newline at end of file diff --git a/.github/workflows/publish_docker.yml b/.github/workflows/publish_docker.yml new file mode 100644 index 00000000..80739507 --- /dev/null +++ b/.github/workflows/publish_docker.yml @@ -0,0 +1,69 @@ +name: publish_docker + +on: + push: + tags: + - v* + branches: + - master + +jobs: + publish_docker: + runs-on: ubuntu-latest + + permissions: + id-token: write + contents: read + env: + DOCKER_TAG: ${{github.ref_type == 'tag' && github.ref_name \|\| 'latest'}} + REPO_NAME: 'augerai/a2ml' + DOCKER_USER: ${{ secrets.DOCKER_USER }} + DOCKER_PASS: ${{ secrets.DOCKER_PASS }} + + AWS_KS_ROLE: ${{ secrets[format('AWS_KS_{0}_ROLE', github.ref_type == 'tag' && 'STABLE' \|\| 'EXPERIMENTAL')] }} + CLUSTER_NAME: ${{ secrets[format('{0}_CLUSTER_NAME', github.ref_type == 'tag' && 'STABLE' \|\| 'EXPERIMENTAL')] }} + KUBECONFIG_FILE: '/home/runner/.kube/config' + RELEASE_NAME: 'a2ml' + + steps: + - uses: actions/checkout@v3 + + - name: Build docker and run tests + run: make config docker-test + + - name: Push docker image + run: \| + docker login -u $DOCKER_USER -p $DOCKER_PASS + docker push $REPO_NAME:$DOCKER_TAG + + - name: configure aws credentials with role1 + uses: aws-actions/configure-aws-credentials@v1 + with: + role-to-assume: ${{ secrets.AWS_ROLE }} + aws-region: us-west-2 + + - name: Assume execution role + uses: aws-actions/configure-aws-credentials@v1 + with: + aws-access-key-id: ${{ env.AWS_ACCESS_KEY_ID }} + aws-region: us-west-2 + aws-secret-access-key: ${{ env.AWS_SECRET_ACCESS_KEY }} + aws-session-token: ${{ env.AWS_SESSION_TOKEN }} + role-duration-seconds: 3000 + role-skip-session-tagging: true + role-to-assume: ${{ env.AWS_KS_ROLE }} + + - uses: azure/[email protected] + - uses: azure/setup-helm@v1 + with: + version: 'v3.4.2' + + - name: setup the kubectl config + run : aws eks update-kubeconfig --name $CLUSTER_NAME + + - name: Helm upgrade augerai repo + run: \| + helm repo add augerai https://augerai.github.io/charts + helm repo update + helm repo list + helm upgrade $RELEASE_NAME $REPO_NAME --namespace=$RELEASE_NAME --set=image.tag=$DOCKER_TAG --reuse-values --wait --atomic --kubeconfig $KUBECONFIG_FILE diff --git a/.github/workflows/publish_pip.yml b/.github/workflows/publish_pip.yml new file mode 100644 index 00000000..e10e2f9a --- /dev/null +++ b/.github/workflows/publish_pip.yml @@ -0,0 +1,33 @@ +name: publish_pip + +on: + push: + tags: + - v* + +jobs: + publish_pip: + runs-on: ubuntu-latest + env: + PYPI_USERNAME: ${{ secrets.PYPI_USERNAME }} + PYPI_PASSWORD: ${{ secrets.PYPI_PASSWORD }} + + steps: + - uses: actions/checkout@v3 + - name: Set up Python + uses: actions/setup-python@v3 + with: + python-version: 3.7 + + - name: Make pypirc file + run: \| + echo -e "[pypi]" >> ~/.pypirc + echo -e "username = $PYPI_USERNAME" >> ~/.pypirc + echo -e "password = $PYPI_PASSWORD" >> ~/.pypirc + + - name: Install dependencies + run: pip install wheel + - name: Build wheel + run: make build + - name: Deploy package + run: make release diff --git a/a2ml/__init__.py b/a2ml/__init__.py index 53a7c762..461d9678 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.76' +__version__ = '1.0.77' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index dc72609a..03454e01 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -8,7 +8,7 @@ class A2ML(BaseA2ML): """Facade to A2ML providers.""" def __init__(self, ctx, provider = None): - """Initializes A2ML PREDIT instance. + """Initializes new A2ML PREDIT instance. Args: ctx (object): An instance of the a2ml Context. diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index f84dfa20..11f95335 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -79,6 +79,8 @@ def _do_score_actual(self, df_data, predicted_feature=None, extra_features=[]): def validate_roi_syntax(self, expressions, features=[]): res = [] + logging.info('validate_roi_syntax with experession: %s'%(expressions)) + known_vars = ["A", "P", "$" + self.target_feature] + list( map(lambda name: "$" + name, set(self.original_features + features)) ) diff --git a/docs/requirements.txt b/docs/requirements.txt index 51d880e1..0f8a6e5f 100644 --- a/docs/requirements.txt +++ b/docs/requirements.txt @@ -1,2 +1,2 @@ -sphinx~=3.0.4 +sphinx~=4.4.0 git+https://github.com/augerai/sphinx_rtd_theme.git@bump-version#2ab38df0d303163e0e6c2bac80d907e9915000cb'	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2022-05-09T20:42:48	[]	[]
augerai/a2ml	augerai__a2ml-607	2273c754ddc486783b6b1095e15c5b67e7c48242	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index a53afbe9..218e3516 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.65' +__version__ = '1.0.66' diff --git a/a2ml/api/auger/config.py b/a2ml/api/auger/config.py index 0a9e47a5..3d8013fb 100644 --- a/a2ml/api/auger/config.py +++ b/a2ml/api/auger/config.py @@ -3,7 +3,7 @@ def __init__(self, ctx): super(AugerConfig, self).__init__() self.ctx = ctx - def set_data_set(self, name, source=None, validation=False): + def set_data_set(self, name, source=None, validation=False, user_name=None): #TODO: add more providers later if validation: self.ctx.config.set('experiment/validation_dataset', name) @@ -12,6 +12,9 @@ def set_data_set(self, name, source=None, validation=False): else: #print("set_data_set: %s"%self.ctx.use_auger_cloud()) self.ctx.config.set('dataset', name) + if user_name: + self.ctx.config.set('dataset_name', user_name) + if self.ctx.use_auger_cloud() and 'azure' in self.ctx.get_providers(): self.ctx.config.set('dataset', name, "azure") @@ -21,9 +24,34 @@ def set_data_set(self, name, source=None, validation=False): def set_experiment(self, experiment_name, experiment_session_id): self.ctx.config.set('experiment/name', experiment_name) self.ctx.config.set('experiment/experiment_session_id', experiment_session_id) + + if self.ctx.config.get('dataset_name'): + dataset_name = self.ctx.config.get('dataset_name') + self.ctx.config.set(f'experiments/{dataset_name}/experiment_id', experiment_name) + self.ctx.config.set(f'experiments/{dataset_name}/experiment_session_id', experiment_session_id) + self.ctx.config.write() + def _get_experiment_by_dataset(self): + dataset_name = self.ctx.config.get('dataset_name') + experiments = self.ctx.config.get('experiments', {}) + + return experiments.get(dataset_name, {}) + + def get_experiment(self): + return self._get_experiment_by_dataset().get('experiment_id', + self.ctx.config.get('experiment/name')) + + def get_experiment_session(self): + return self._get_experiment_by_dataset().get('experiment_session_id', + self.ctx.config.get('experiment/experiment_session_id')) + + def get_dataset(self): + return self._get_experiment_by_dataset().get('dataset_id', + self.ctx.config.get('dataset')) + def set_project(self, project_name): self.ctx.config.set('name', project_name) self.ctx.config.write() return self + diff --git a/a2ml/api/auger/dataset.py b/a2ml/api/auger/dataset.py index 65ec91ea..a99cc098 100644 --- a/a2ml/api/auger/dataset.py +++ b/a2ml/api/auger/dataset.py @@ -18,7 +18,7 @@ def __init__(self, ctx): @with_project(autocreate=False) def list(self, project): count = 0 - selected = self.ctx.config.get('dataset', None) + selected = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) for dataset in iter(DataSet(self.ctx, project).list()): self.ctx.log( ('[%s] ' % ('x' if selected == dataset.get('name') else ' ')) + @@ -40,7 +40,7 @@ def _create(self, project, source = None, validation=False, name=None, descripti if source is None: source = self.ctx.config.get('source', None) dataset = DataSet(self.ctx, project).create(source, name, description) - AugerConfig(self.ctx).set_data_set(dataset.name, source, validation) + AugerConfig(self.ctx).set_data_set(dataset.name, source, validation, name) return dataset @@ -60,7 +60,7 @@ def upload(self, project, source = None, name=None): @with_project(autocreate=False) def delete(self, project, name): if name is None: - name = self.ctx.config.get('dataset', None) + name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) DataSet(self.ctx, project, name).delete() if name == self.ctx.config.get('dataset', None): AugerConfig(self.ctx).set_data_set(None, None, False).set_experiment(None, None) @@ -69,7 +69,7 @@ def delete(self, project, name): @error_handler def select(self, name): - old_name = self.ctx.config.get('dataset', None) + old_name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) if name != old_name: AugerConfig(self.ctx).set_data_set(name, None, False).set_experiment(None, None) self.ctx.log('Selected DataSet %s' % name) @@ -80,7 +80,7 @@ def select(self, name): @with_project(autocreate=False) def download(self, project, name, path_to_download): if name is None: - name = self.ctx.config.get('dataset', None) + name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) file_name = DataSet(self.ctx, project, name).download(path_to_download) self.ctx.log('Downloaded dataset %s to %s' % (name, file_name)) return {'dowloaded': name, 'file': file_name} diff --git a/a2ml/api/auger/experiment.py b/a2ml/api/auger/experiment.py index edb4df07..04dbc3cc 100644 --- a/a2ml/api/auger/experiment.py +++ b/a2ml/api/auger/experiment.py @@ -31,7 +31,7 @@ def list(self, dataset): @with_dataset def start(self, dataset): experiment_name = \ - self.ctx.config.get('experiment/name', None) + AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) experiment_name, session_id = \ Experiment(self.ctx, dataset, experiment_name).start() AugerConfig(self.ctx).set_experiment(experiment_name, session_id) @@ -41,12 +41,12 @@ def start(self, dataset): @authenticated @with_dataset def stop(self, dataset, run_id = None): - name = self.ctx.config.get('experiment/name', None) + name = AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) if name is None: raise AugerException('Please specify Experiment name...') if run_id is None: - run_id = self.ctx.config.get( - 'experiment/experiment_session_id', None) + run_id = AugerConfig(self.ctx).get_experiment_session() + #self.ctx.config.get('experiment/experiment_session_id', None) if Experiment(self.ctx, dataset, name).stop(run_id): self.ctx.log('Search is stopped...') @@ -58,12 +58,13 @@ def stop(self, dataset, run_id = None): @authenticated @with_dataset def leaderboard(self, dataset, run_id = None): - name = self.ctx.config.get('experiment/name', None) + name = AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) if name is None: raise AugerException('Please specify Experiment name...') if run_id is None: - run_id = self.ctx.config.get( - 'experiment/experiment_session_id', None) + run_id = AugerConfig(self.ctx).get_experiment_session() + # run_id = self.ctx.config.get( + # 'experiment/experiment_session_id', None) leaderboard, status, run_id, trials_count, errors = Experiment( self.ctx, dataset, name).leaderboard(run_id) if leaderboard is None: @@ -109,7 +110,7 @@ def leaderboard(self, dataset, run_id = None): @authenticated @with_dataset def history(self, dataset): - name = self.ctx.config.get('experiment/name', None) + name = AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) if name is None: raise AugerException('Please specify Experiment name...') for exp_run in iter(Experiment(self.ctx, dataset, name).history()): diff --git a/a2ml/api/auger/impl/decorators.py b/a2ml/api/auger/impl/decorators.py index aaa73523..f681cad2 100644 --- a/a2ml/api/auger/impl/decorators.py +++ b/a2ml/api/auger/impl/decorators.py @@ -33,10 +33,11 @@ def wrapper(self, args, kwargs): def with_dataset(decorated): from .dataset import DataSet + from ..config import AugerConfig def wrapper(self, args, *kwargs): project = _get_project(self, False) - data_set_name = self.ctx.config.get('dataset', None) + data_set_name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) if data_set_name is None: raise AugerException( 'Please specify dataset name in auger.yaml/dataset...') diff --git a/a2ml/api/utils/context.py b/a2ml/api/utils/context.py index 853aa570..cab0cf6c 100644 --- a/a2ml/api/utils/context.py +++ b/a2ml/api/utils/context.py @@ -14,7 +14,7 @@ class Context(object): """The Context class provides an environment to run A2ML""" - def __init__(self, name='config', path=None, debug=False): + def __init__(self, name='auger', path=None, debug=False): """Initializes the Context instance Args: @@ -39,7 +39,7 @@ def __init__(self, name='config', path=None, debug=False): self.provider_info = None if len(self.name) > 0: - self.name = "{:<9}".format('[%s]' % self.name) + self.name = f'[{self.name}] ' #"{:<9}".format('[%s]' % self.name) self.debug = self.config.get('debug', debug) self.set_runs_on_server(False) @@ -109,6 +109,8 @@ def is_external_provider(self): return providers and providers[0] == 'external' def copy(self, name): + return self + """creates a copy of an existing Context Args: @@ -123,23 +125,23 @@ def copy(self, name): ctx = Context() new_ctx = ctx.copy() """ - new = Context(name, self.config.path, self.debug) - new.set_runs_on_server(self._runs_on_server) - new.notificator = self.notificator - new.request_id = self.request_id - new.config.parts = self.config.parts - new.config.parts_changes = self.config.parts_changes - - try: - new.config.set("providers", name, config_name='config') - except Exception as e: - # In case if command run in folder without config, do not set it - pass + # new = Context(name, self.config.path, self.debug) + # new.set_runs_on_server(self._runs_on_server) + # new.notificator = self.notificator + # new.request_id = self.request_id + # new.config.parts = self.config.parts + # new.config.parts_changes = self.config.parts_changes + + # try: + # new.config.set("providers", name, config_name='config') + # except Exception as e: + # # In case if command run in folder without config, do not set it + # pass - if hasattr(self, 'credentials'): - new.credentials = self.credentials + # if hasattr(self, 'credentials'): + # new.credentials = self.credentials - return new + # return new def log(self, msg, args, *kwargs): log.info('%s%s' %(self.name, msg), args, **kwargs) diff --git a/a2ml/cmdl/commands/cmd_import.py b/a2ml/cmdl/commands/cmd_import.py index 2502ef75..fcc87dbc 100644 --- a/a2ml/cmdl/commands/cmd_import.py +++ b/a2ml/cmdl/commands/cmd_import.py @@ -5,12 +5,14 @@ @click.command('import', short_help='Import data for training.') @click.option('--source', '-s', type=click.STRING, required=False, help='Source file to import.If skipped, then import source from config.yml.') [email protected]('--name', '-n', type=click.STRING, required=False, + help='Name file to import.') @click.option('--description', '-d', type=click.STRING, required=False, help='Description of dataset.') @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') @pass_context -def cmdl(ctx, source, description, provider): +def cmdl(ctx, source, name, description, provider): """Import data for training.""" ctx.setup_logger(format='') - A2ML(ctx, provider).import_data(source, description=description) + A2ML(ctx, provider).import_data(source, name, description=description)	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2022-03-17T09:51:24	[]	[]
augerai/a2ml	augerai__a2ml-601	bb141e98e7cb89a5d6aa226cb98adbb1b3163add	diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 69f5b188..8424c81a 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -365,7 +365,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at score, score_true_data ) @show_result - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False, provider=None): """Submits actual results(ground truths) for predictions of a deployed model. This is used to review and monitor active models. Note: @@ -396,6 +397,12 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N columns(list): list of column names if data is array of records actuals_at: Actuals date. Use for review of historical data. actual_date_column(str): name of column in data which contains actual date + experiment_params(dict): parameters to calculate experiment metrics :: + + start_date(date): experiment actuals start date + end_date(date): experiment actuals end date + date_col(str): column name with date + locally(bool): Process actuals locally. provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. @@ -437,7 +444,8 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N A2ML(ctx, "external").actuals('external_model_id', data=actual_records,columns=columns) """ - return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', model_id, filename, data, columns, actuals_at, actual_date_column, locally) + return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', + model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) @show_result diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index aaf1cf0c..86dae92b 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -155,7 +155,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at score, score_true_data ) @show_result - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False, provider=None): """Submits actual results(ground truths) for predictions of a deployed model. This is used to review and monitor active models. Note: @@ -186,6 +187,13 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N columns(list): list of column names if data is array of records actuals_at: Actuals date. Use for review of historical data. actual_date_column(str): name of column in data which contains actual date + experiment_params(dict): parameters to calculate experiment metrics :: + + start_date(date): experiment actuals start date + end_date(date): experiment actuals end date + date_col(str): column name with date + + locally(bool): Process actuals locally. provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. @@ -227,7 +235,8 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N A2MLModel(ctx, "external").actuals('external_model_id', data=actual_records,columns=columns) """ - return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', model_id, filename, data, columns, actuals_at, actual_date_column, locally) + return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', + model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) @show_result def review_alert(self, model_id, parameters = None, locally=False, provider=None, name=None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 27c75909..46f7db9d 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -30,9 +30,10 @@ def predict(self, model_id, filename, threshold=None, locally=False, data=None, model_id, filename, threshold, locally, data, columns, predicted_at, output, no_features_in_result, score, score_true_data) - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False): return AugerModel(self.ctx).actuals( - model_id, filename, data, columns, actuals_at, actual_date_column, locally) + model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) def delete_actuals(self, model_id, with_predictions=False, begin_date=None, end_date=None, locally=False): return AugerModel(self.ctx).delete_actuals( diff --git a/a2ml/api/auger/impl/cloud/actual.py b/a2ml/api/auger/impl/cloud/actual.py index 17e08373..c29c7f23 100644 --- a/a2ml/api/auger/impl/cloud/actual.py +++ b/a2ml/api/auger/impl/cloud/actual.py @@ -12,7 +12,7 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self.parent_id_name = "endpoint_id" self._set_api_request_path("AugerEndpointActualApi") - def create(self, records, features, actuals_at, actuals_path, actual_date_column): + def create(self, records, features, actuals_at, actuals_path, actual_date_column, experiment_params): params = {} if self.use_endpoint: params['endpoint_id'] = self.parent_api.object_id @@ -35,6 +35,10 @@ def create(self, records, features, actuals_at, actuals_path, actual_date_column if actuals_at: params['actuals_at'] = str(actuals_at) + if experiment_params: + params['experiment_params'] = experiment_params + + print(params) return self._call_create( params=params, has_return_object=False) diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index 4c2e4605..b5c1847e 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -94,12 +94,12 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at return prediction_properties.get('result') - def actual(self, records, features, actuals_at, actuals_path, actual_date_column): + def actual(self, records, features, actuals_at, actuals_path, actual_date_column, experiment_params): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') actual_api = AugerActualApi(self.ctx, self, use_endpoint=self.check_endpoint()) - actual_api.create(records, features, actuals_at, actuals_path, actual_date_column) + actual_api.create(records, features, actuals_at, actuals_path, actual_date_column, experiment_params) #TODO: get object actual from cloud return True diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 89e2b25e..72826d08 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -35,7 +35,8 @@ def predict(self, filename, model_id, threshold=None, locally=False, data=None, return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result, score, score_true_data) - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False): if locally: is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) @@ -57,10 +58,12 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N columns=columns, actual_date=actuals_at, actual_date_column=actual_date_column, + experiment_params=experiment_params, provider='auger' ) else: - return ModelActual(self.ctx).execute(model_id, filename, data, columns, actuals_at, actual_date_column) + return ModelActual(self.ctx).execute(model_id, filename, data, columns, actuals_at, + actual_date_column, experiment_params) def delete_actuals(self, model_id, with_predictions=False, begin_date=None, end_date=None, locally=False): if locally: diff --git a/a2ml/api/auger/impl/mparts/actual.py b/a2ml/api/auger/impl/mparts/actual.py index aad79269..50edcb7b 100644 --- a/a2ml/api/auger/impl/mparts/actual.py +++ b/a2ml/api/auger/impl/mparts/actual.py @@ -12,9 +12,9 @@ def __init__(self, ctx): super(ModelActual, self).__init__() self.ctx = ctx - def execute(self, model_id, filename, data, columns, actuals_at, actual_date_column): + def execute(self, model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params): records, features, file_url, is_pandas_df = ModelPredict(self.ctx)._process_input(filename, data, columns=columns) pipeline_api = AugerPipelineApi(self.ctx, None, model_id) - return pipeline_api.actual(records, features, actuals_at, file_url, actual_date_column) + return pipeline_api.actual(records, features, actuals_at, file_url, actual_date_column, experiment_params) diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 7901f871..326a219f 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -42,8 +42,9 @@ def predict(self, filename, model_id, threshold, locally, data, columns, predict @error_handler @authenticated @with_project(autocreate=False) - def actuals(self, project, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): - return Model(self.ctx, project).actuals(model_id, filename, data, columns, actuals_at, actual_date_column, locally) + def actuals(self, project, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False): + return Model(self.ctx, project).actuals(model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) @error_handler @authenticated diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index 6df5be05..1d2e0bf9 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -145,7 +145,7 @@ def add_external_model(self, target_column, scoring, task_type, binary_classific def add_actuals( self, ctx, actuals_path=None, data=None, columns=None, external_model=False, actual_date=None, actual_date_column=None, actuals_id = None, return_count=False, provider='auger', - do_predict=False + do_predict=False, experiment_params=None ): ds_actuals = DataFrame.create_dataframe(actuals_path, data, features=columns) @@ -178,8 +178,12 @@ def add_actuals( result = self._do_score_actual(ds_actuals.df) baseline_score = {} + experiment_score = {} + experiment_count = 0 if "baseline_target" in ds_actuals.columns: baseline_score = self._do_score_actual(ds_actuals.df, "baseline_target") + if experiment_params: + experiment_score, experiment_count = self._do_score_actual_experiment(ds_actuals, experiment_params) #logging.info("Actual result: %s", result) ds_actuals.df = ds_actuals.df.rename(columns={self.target_feature: 'a2ml_predicted'}) @@ -207,10 +211,34 @@ def add_actuals( ds_actuals.saveToFeatherFile(os.path.join(self.model_path, "predictions", file_name)) if return_count: - return {'score': result, 'count': actuals_count, 'baseline_score': baseline_score} + return {'score': result, 'count': actuals_count, 'baseline_score': baseline_score, + 'experiment_score': experiment_score, 'experiment_count': experiment_count} else: return result + def _do_score_actual_experiment(self, ds_actuals, experiment_params): + if experiment_params.get('start_date') and experiment_params.get('end_date'): + df_exp_actuals = ds_actuals.df.query("%s>='%s' and %s<'%s'"%( + experiment_params.get('date_col'), + experiment_params.get('start_date'), + experiment_params.get('date_col'), + experiment_params.get('end_date') + )) + elif experiment_params.get('start_date'): + df_exp_actuals = ds_actuals.df.query("%s>='%s'"%( + experiment_params.get('date_col'), + experiment_params.get('start_date') + )) + elif experiment_params.get('end_date'): + df_exp_actuals = ds_actuals.df.query("%s<'%s'"%( + experiment_params.get('date_col'), + experiment_params.get('end_date') + )) + else: + df_exp_actuals = ds_actuals.df + + return self._do_score_actual(df_exp_actuals), len(df_exp_actuals) + def _do_predict(self, ctx, ds_actuals, provider, predict_feature=None, predicted_at=None): missing_features = set(self.original_features) - set(ds_actuals.columns) if len(missing_features) > 0: diff --git a/a2ml/tasks_queue/tasks_hub_api.py b/a2ml/tasks_queue/tasks_hub_api.py index 73bafa16..d5ccbae4 100644 --- a/a2ml/tasks_queue/tasks_hub_api.py +++ b/a2ml/tasks_queue/tasks_hub_api.py @@ -560,7 +560,8 @@ def score_actuals_by_model_task(params): return_count=params.get('return_count', False), provider=params.get('provider'), external_model=external_model, - do_predict=params.get('do_predict', False) + do_predict=params.get('do_predict', False), + experiment_params=params.get('experiment_params'), ) @celeryApp.task(ignore_result=True)	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-10-29T17:51:10	[]	[]
augerai/a2ml	augerai__a2ml-594	c255bd05603d41d4bfb604ec940225d2656b0b6c	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index b41bc4d5..f47e40c5 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.37' +__version__ = '1.0.38' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 14fd78f0..6049297e 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -1,6 +1,6 @@ from a2ml.api.base_a2ml import BaseA2ML from a2ml.api.utils.show_result import show_result -from a2ml.api.utils import convert_source +#from a2ml.api.utils import convert_source from a2ml.api.utils.context import Context @@ -28,7 +28,7 @@ def __init__(self, ctx, provider = None): self.local_runner = lambda: self.build_runner(ctx, provider, force_local=True) @show_result - def import_data(self, source=None): + def import_data(self, source=None, name=None): """Imports data defined in context. Uploading the same file name will result in versions being appended to the file name. Note: @@ -46,6 +46,7 @@ def import_data(self, source=None): Args: source (str, optional): Local file name, remote url to the data source file, Pandas DataFrame or postgres url + name (str, optional): Name of dataset, if none then file name used. If source is DataFrame then name should be specified. Returns: Results for each provider. :: @@ -70,8 +71,7 @@ def import_data(self, source=None): a2ml = A2ML(ctx, 'auger, azure') a2ml.import_data() """ - with convert_source(source, self.ctx.config.get("name", "source_data")) as data_source: - return self.runner.execute('import_data', source=data_source) + return self.runner.execute('import_data', source=source, name=name) @show_result def preprocess_data(self, data, preprocessors, locally=False): diff --git a/a2ml/api/a2ml_dataset.py b/a2ml/api/a2ml_dataset.py index 1a623e87..9861efa2 100644 --- a/a2ml/api/a2ml_dataset.py +++ b/a2ml/api/a2ml_dataset.py @@ -1,6 +1,6 @@ from a2ml.api.base_a2ml import BaseA2ML from a2ml.api.utils.show_result import show_result -from a2ml.api.utils import convert_source +#from a2ml.api.utils import convert_source class A2MLDataset(BaseA2ML): @@ -58,11 +58,12 @@ def list(self): return self.runner.execute('list') @show_result - def create(self, source = None): + def create(self, source = None, name=None): """Create a new DataSet for the Project specified in the .yaml. Args: source (str, optional): Local file name, remote url to the data source file, Pandas DataFrame or postgres url + name (str, optional): Name of dataset, if none then file name used. If source is DataFrame then name should be specified. Returns: Results for each provider. :: @@ -82,8 +83,7 @@ def create(self, source = None): ctx = Context() dataset = DataSet(ctx, 'auger, azure').create('../dataset.csv') """ - with convert_source(source, self.ctx.config.get("name", "source_data")) as data_source: - return self.runner.execute('create', data_source) + return self.runner.execute('create', data_source, name) @show_result def delete(self, name = None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 394c873f..86cc3b80 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -9,8 +9,8 @@ def __init__(self, ctx): super(AugerA2ML, self).__init__() self.ctx = ctx - def import_data(self, source=None): - return AugerDataset(self.ctx).create(source=source) + def import_data(self, source=None, name=None): + return AugerDataset(self.ctx).create(source=source, name=name) def preprocess_data(self, data, preprocessors, locally=False): return AugerDataset(self.ctx).preprocess_data(data, preprocessors, locally) diff --git a/a2ml/api/auger/dataset.py b/a2ml/api/auger/dataset.py index 4e7ffc6a..0619f62c 100644 --- a/a2ml/api/auger/dataset.py +++ b/a2ml/api/auger/dataset.py @@ -31,15 +31,15 @@ def list(self, project): @error_handler @authenticated @with_project(autocreate=True) - def create(self, project, source = None, validation=False): - dataset = self._create(project, source, validation) + def create(self, project, source = None, validation=False, name=None): + dataset = self._create(project, source, validation, name) self.ctx.log('Created DataSet %s' % dataset.name) return {'created': dataset.name} - def _create(self, project, source = None, validation=False): + def _create(self, project, source = None, validation=False, name=None): if source is None: source = self.ctx.config.get('source', None) - dataset = DataSet(self.ctx, project).create(source) + dataset = DataSet(self.ctx, project).create(source, name) AugerConfig(self.ctx).set_data_set(dataset.name, source, validation) return dataset diff --git a/a2ml/api/auger/impl/cloud/dataset.py b/a2ml/api/auger/impl/cloud/dataset.py index 6419bfda..ae56a0c7 100644 --- a/a2ml/api/auger/impl/cloud/dataset.py +++ b/a2ml/api/auger/impl/cloud/dataset.py @@ -6,6 +6,7 @@ import urllib.parse import urllib.request import xml.etree.ElementTree as ElementTree +from a2ml.api.utils.dataframe import DataFrame from .cluster import AugerClusterApi from .project_file import AugerProjectFileApi @@ -31,8 +32,17 @@ def do_upload_file(self, data_source_file, data_set_name=None, local_data_source # AugerDataSetApi.verify(data_source_file, self.ctx.config.path) if local_data_source: - file_url = self._upload_to_cloud(data_source_file) - file_name = os.path.basename(data_source_file) + if DataFrame.is_dataframe(data_source_file): + with fsclient.save_atomic("%s.parquet"%data_set_name, move_file=False) as local_path: + ds = DataFrame.create_dataframe(data_source_file) + ds.saveToParquetFile(local_path) + file_url = self._upload_to_cloud(local_path) + + file_name = data_set_name + else: + file_url = self._upload_to_cloud(data_source_file) + file_name = os.path.basename(data_source_file) + if data_set_name: self.object_name = data_set_name else: @@ -81,6 +91,9 @@ def _get_readable_name(self): @staticmethod def verify(data_source_file, config_path=None): + if DataFrame.is_dataframe(data_source_file): + return data_source_file, True + if urllib.parse.urlparse(data_source_file).scheme in ['http', 'https']: return data_source_file, False diff --git a/a2ml/api/auger/impl/dataset.py b/a2ml/api/auger/impl/dataset.py index 6a86af1a..b05fb435 100644 --- a/a2ml/api/auger/impl/dataset.py +++ b/a2ml/api/auger/impl/dataset.py @@ -10,7 +10,7 @@ def __init__(self, ctx, project, data_set_name=None): ctx, project, data_set_name) self.project = project - def create(self, data_source_file): + def create(self, data_source_file, name): if data_source_file is None: raise AugerException('Please specify data source file...') @@ -19,7 +19,7 @@ def create(self, data_source_file): self.project.start() - super().create(data_source_file, self.object_name, local_data_source=local_data_source) + super().create(data_source_file, name if name else self.object_name, local_data_source=local_data_source) return self def upload_file(self, data_source_file): diff --git a/a2ml/api/utils/__init__.py b/a2ml/api/utils/__init__.py index 22429b26..6acf1a52 100644 --- a/a2ml/api/utils/__init__.py +++ b/a2ml/api/utils/__init__.py @@ -215,14 +215,14 @@ def convert_to_date(date): else: return date [email protected] -def convert_source(source, name): - if source is not None and isinstance(source, pd.DataFrame): - with fsclient.save_atomic("%s.parquet"%name, move_file=False) as local_path: - source.to_parquet(local_path, index=False, compression="gzip") - yield local_path - else: - yield source +# @contextlib.contextmanager +# def convert_source(source, name): +# if source is not None and isinstance(source, pd.DataFrame): +# with fsclient.save_atomic("%s.parquet"%name, move_file=False) as local_path: +# source.to_parquet(local_path, index=False, compression="gzip") +# yield local_path +# else: +# yield source def retry_helper(func, retry_errors=[], num_try=10, delay=10, ctx=None): nTry = 0 diff --git a/a2ml/api/utils/dataframe.py b/a2ml/api/utils/dataframe.py index ca737052..d81998f7 100644 --- a/a2ml/api/utils/dataframe.py +++ b/a2ml/api/utils/dataframe.py @@ -43,21 +43,48 @@ def _get_compression(self, extension): return compression @staticmethod - def create_dataframe(data_path=None, records=None, features=None): - if data_path: - ds = DataFrame({'data_path': data_path}) - ds.load(features = features) - elif records is not None and isinstance(records, pd.DataFrame): - ds = DataFrame({}) - ds.df = records - if features: - ds.df = ds.df[features] - - ds.from_pandas = True + def create_dataframe(data_path=None, records=None, features=None, reset_index=False): + if data_path is not None: + if isinstance(data_path, pd.DataFrame): + ds = DataFrame({}) + ds.df = data_path + elif isinstance(data_path, DataFrame): + ds = data_path + elif isinstance(data_path, list): + ds = DataFrame({}) + ds.load_records(data_path) + elif isinstance(data_path, dict): + ds = DataFrame({}) + + if 'data' in data_path and 'columns' in data_path: + ds.load_records(data_path['data'], features=data_path['columns']) + else: + ds.load_records(data_path) + else: + ds = DataFrame({'data_path': data_path}) + ds.load(features = features) else: ds = DataFrame({}) ds.load_records(records, features=features) + if reset_index and ds.df is not None: + ds.df.reset_index(drop=True, inplace=True) + + + # if data_path: + # ds = DataFrame({'data_path': data_path}) + # ds.load(features = features) + # elif records is not None and isinstance(records, pd.DataFrame): + # ds = DataFrame({}) + # ds.df = records + # if features: + # ds.df = ds.df[features] + + # ds.from_pandas = True + # else: + # ds = DataFrame({}) + # ds.load_records(records, features=features) + return ds @staticmethod @@ -72,6 +99,10 @@ def load_from_files(files, features=None): except Exception as exc: logging.exception("load_from_files failed for: %s. Error: %s"%(path, exc)) + @staticmethod + def is_dataframe(data): + return isinstance(data, pd.DataFrame) or isinstance(data, DataFrame) + def load_from_file(self, path, features=None, nrows=None): from collections import OrderedDict	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-09-01T21:01:12	[]	[]
augerai/a2ml	augerai__a2ml-591	e8fc37867e2ede621fa14a6877f661d62de5ae20	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index 1c327d42..24d5b4c5 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.34' +__version__ = '1.0.35' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index aa0e4284..14fd78f0 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -73,6 +73,61 @@ def import_data(self, source=None): with convert_source(source, self.ctx.config.get("name", "source_data")) as data_source: return self.runner.execute('import_data', source=data_source) + @show_result + def preprocess_data(self, data, preprocessors, locally=False): + """Preprocess data + + Args: + data (str\|pandas.DataFrame): Input data for preprocess. Can be path to file(local or s3) or Pandas Dataframe + preprocessors (array of dicts): List of preprocessors with parameters :: + + [ + {'text': {'text_cols': []}} + ] + + Preprocessors: + text + * text_cols(array): List of text columns to process + * text_metrics ['mean_length', 'unique_count', 'separation_score'] : Calculate metrics for text fields and after vectorize(separation_score) + * tokenize (dict): Default - {'max_text_len': 30000, 'tokenizers': ['sent'], 'remove_chars': '○•'} + * vectorize ('en_use_lg'\|'hashing'\|'en_use_md'\|'en_use_cmlm_md'\|'en_use_cmlm_lg'): See see https://github.com/MartinoMensio/spacy-universal-sentence-encoder + * dim_reduction(dict): Generate features based on vectors. See https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html :: + + { + 'alg_name': 'PCA'\|'t-SNE', + 'args': {'n_components': 2} #Number of components to keep. + } + + * output_prefix (str): Prefix for generated columns. Format name: {prefix}_{colname}_{num} + + * calc_distance ['none', 'cosine', 'cityblock', 'euclidean', 'haversine', 'l1', 'l2', 'manhattan', 'nan_euclidean'] \| 'cosine' : See https://scikit-learn.org/stable/modules/generated/sklearn.metrics.pairwise.distance_metrics.html#sklearn.metrics.pairwise.distance_metrics + * compare_pairs (array of dicts): When calc_distance is not none. :: + + [ + {'compare_cols': [{'dataset_idx': 0, 'cols': ['col1']}, {'dataset_idx': 1, 'cols': ['col2']}], + 'output_name':'cosine_col1_col2', 'params': {} + }, + {'compare_cols': [{'dataset_idx': 0, 'cols': ['col3']}, {'dataset_idx': 1, 'cols': ['col4']}], + 'output_name':'cosine_col3_col4', 'params': {} + }, + ] + + * datasets: List of datasets to process, may be empty, so all fields takes from main dataset :: + + [ + {'path': 'path', 'keys': ['main_key', 'local_key'], 'text_metrics': ['separation_score', 'mean_length', 'unique_count']}, + {'path': 'path1', 'keys': ['main_key1', 'local_key1']} + ] + + Returns: + { + 'result': True, + 'data': 'data in input format' + } + + """ + return self.get_runner(locally).execute_one_provider('preprocess_data', data, preprocessors, locally) + @show_result def train(self): """Starts training session based on context state. diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 2a55df5e..394c873f 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -12,6 +12,9 @@ def __init__(self, ctx): def import_data(self, source=None): return AugerDataset(self.ctx).create(source=source) + def preprocess_data(self, data, preprocessors, locally=False): + return AugerDataset(self.ctx).preprocess_data(data, preprocessors, locally) + def train(self): return AugerExperiment(self.ctx).start() diff --git a/a2ml/api/auger/dataset.py b/a2ml/api/auger/dataset.py index dc0b6fee..4e7ffc6a 100644 --- a/a2ml/api/auger/dataset.py +++ b/a2ml/api/auger/dataset.py @@ -73,3 +73,25 @@ def download(self, project, name, path_to_download): file_name = DataSet(self.ctx, project, name).download(path_to_download) self.ctx.log('Downloaded dataset %s to %s' % (name, file_name)) return {'dowloaded': name, 'file': file_name} + + def preprocess_data(self, data, preprocessors, locally): + if locally: + return self._preprocess_data_locally(data, preprocessors) + else: + raise Exception("preprocess_data supported with locally=True only.") + + def _preprocess_data_locally(self, data, preprocessors): + from auger_ml.preprocessors.text import TextPreprocessor + + res = data + for p in preprocessors: + name = list(p.keys())[0] + params = list(p.values())[0] + if name != 'text': + raise Exception("Only text preprocessor supported.") + + tp = TextPreprocessor(params) + res = tp.fit_transform(res) + + return res + \ No newline at end of file diff --git a/setup.py b/setup.py index 0a6f572f..2b11a88c 100644 --- a/setup.py +++ b/setup.py @@ -86,7 +86,7 @@ def run(self): 'google-cloud-automl' ], 'predict': [ - 'auger.ai.predict==1.0.79' + 'auger.ai.predict==1.0.80' ] }	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-08-12T20:59:36	[]	[]
augerai/a2ml	augerai__a2ml-587	858d43f5475354996519821e2f99d8a038440a23	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index 93dae163..0630b51e 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.32' +__version__ = '1.0.33' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 09fe4116..aa0e4284 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -221,7 +221,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, no_features_in_result = None, locally=False, provider=None): + threshold=None, score=False, score_true_data=None, + output=None, no_features_in_result = None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -235,6 +236,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at columns(list): list of column names if data is array of records predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. + score(bool): Calculate scores for predicted results. + score_true_data(str, pandas.DataFrame, dict): Data with true values to calculate scores. If missed, target from filename used for true values. output(str): Output csv file path. no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model @@ -301,7 +304,9 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are stored in rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, + threshold, locally, data, columns, predicted_at, output, no_features_in_result, + score, score_true_data ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 2be596f0..b58e2132 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -66,7 +66,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, no_features_in_result=None, locally=False, provider=None): + threshold=None, score=False, score_true_data=None, + output=None, no_features_in_result=None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -80,6 +81,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at columns(list): list of column names if data is array of records predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. + score(bool): Calculate scores for predicted results. + score_true_data(str, pandas.DataFrame, dict): Data with true values to calculate scores. If missed, target from filename used for true values. output(str): Output csv file path. no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model @@ -146,7 +149,9 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are stored in rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, + threshold, locally, data, columns, predicted_at, output, no_features_in_result, + score, score_true_data ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 3e998b5c..2a55df5e 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -21,9 +21,11 @@ def evaluate(self, run_id = None): def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None ): return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) - def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): + def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, + predicted_at=None, output=None, no_features_in_result=None, score=False, score_true_data=None): return AugerModel(self.ctx).predict( - model_id, filename, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + model_id, filename, threshold, locally, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): return AugerModel(self.ctx).actuals( diff --git a/a2ml/api/auger/config.py b/a2ml/api/auger/config.py index 5331a4f8..0a9e47a5 100644 --- a/a2ml/api/auger/config.py +++ b/a2ml/api/auger/config.py @@ -7,12 +7,12 @@ def set_data_set(self, name, source=None, validation=False): #TODO: add more providers later if validation: self.ctx.config.set('experiment/validation_dataset', name) - if self.ctx.use_auger_cloud(): + if self.ctx.use_auger_cloud() and 'azure' in self.ctx.get_providers(): self.ctx.config.set('experiment/validation_dataset', name, "azure") else: - print("set_data_set: %s"%self.ctx.use_auger_cloud()) + #print("set_data_set: %s"%self.ctx.use_auger_cloud()) self.ctx.config.set('dataset', name) - if self.ctx.use_auger_cloud(): + if self.ctx.use_auger_cloud() and 'azure' in self.ctx.get_providers(): self.ctx.config.set('dataset', name, "azure") self.ctx.config.write_all() diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index e3232201..42057f2b 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -49,7 +49,8 @@ def check_endpoint(self, props=None): return is_endpoint - def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): + def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, + no_features_in_result=None, score=False, score_true_data=None): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') @@ -73,8 +74,9 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at prediction_api = AugerPredictionApi(self.ctx, self, use_endpoint=self.check_endpoint(props)) prediction_properties = \ prediction_api.create(records, features, threshold=threshold, file_url=file_url, - predicted_at=predicted_at, no_features_in_result=no_features_in_result) - + predicted_at=predicted_at, no_features_in_result=no_features_in_result, + score=score, score_true_data=score_true_data) + return prediction_properties.get('result') def actual(self, records, features, actuals_at, actuals_path, actual_date_column): diff --git a/a2ml/api/auger/impl/cloud/prediction.py b/a2ml/api/auger/impl/cloud/prediction.py index c9c76704..81f50387 100644 --- a/a2ml/api/auger/impl/cloud/prediction.py +++ b/a2ml/api/auger/impl/cloud/prediction.py @@ -14,7 +14,8 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self.parent_id_name = "endpoint_id" self._set_api_request_path("AugerEndpointPredictionApi") - def create(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): + def create(self, records, features, threshold=None, file_url=None, predicted_at=None, + no_features_in_result=None, score=False, score_true_data=None): params = { 'records': records, 'features': features, @@ -35,5 +36,11 @@ def create(self, records, features, threshold=None, file_url=None, predicted_at= if no_features_in_result is not None: params['no_features_in_result'] = no_features_in_result - + + if score: + params['score'] = score + + if score_true_data: + params['score_true_data'] = score_true_data + return self._call_create(params, ['requested', 'running']) diff --git a/a2ml/api/auger/impl/cloud/rest_api.py b/a2ml/api/auger/impl/cloud/rest_api.py index 58dcb9a7..fd1f39ae 100644 --- a/a2ml/api/auger/impl/cloud/rest_api.py +++ b/a2ml/api/auger/impl/cloud/rest_api.py @@ -25,11 +25,18 @@ def call_ex(self, method, params={}): if params.get('id') and not method.startswith('create_'): oid = params['id'] del params['id'] - #print(method, oid, params) - return getattr(self.hub_client, method)(oid, params) + print(method, oid, params) + res = getattr(self.hub_client, method)(oid, params) else: - #print(method, params) - return getattr(self.hub_client, method)(params) + if method == 'create_endpoint_prediction' or method == 'create_endpoint_actual': + print(method, params.keys()) + else: + print(method, params) + + res = getattr(self.hub_client, method)(params) + + #print(res) + return res def call(self, method, params={}): result = self.call_ex(method, params) diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index c6afa0e7..eeee9960 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -30,11 +30,13 @@ def review(self, model_id): def undeploy(self, model_id, locally=False): return ModelUndeploy(self.ctx, self.project).execute(model_id, locally) - def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): + def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, + output=None, no_features_in_result=None, score=False, score_true_data=None): if locally: self.deploy(model_id, locally) - return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, + predicted_at, output, no_features_in_result, score, score_true_data) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): if locally: diff --git a/a2ml/api/auger/impl/mparts/predict.py b/a2ml/api/auger/impl/mparts/predict.py index 005de202..ec19a7a2 100644 --- a/a2ml/api/auger/impl/mparts/predict.py +++ b/a2ml/api/auger/impl/mparts/predict.py @@ -23,7 +23,8 @@ def __init__(self, ctx): self.ctx = ctx def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, - predicted_at=None, output=None, no_features_in_result=None): + predicted_at=None, output=None, no_features_in_result=None, + score=False, score_true_data=None): if filename and not (filename.startswith("http:") or filename.startswith("https:")) and\ not fsclient.is_s3_path(filename): self.ctx.log('Predicting on data in %s' % filename) @@ -31,11 +32,14 @@ def execute(self, filename, model_id, threshold=None, locally=False, data=None, if locally: if locally == "docker": - predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) else: - predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) else: - predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) return predicted @@ -98,7 +102,8 @@ def _check_model_project(self, pipeline_api): raise AugerException("Project name: %s in config.yml is different from model project name: %s. Please change name in config.yml."%( self.ctx.config.get('name'), model_project_name)) - def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, + output, no_features_in_result, score, score_true_data): records, features, file_url, is_pandas_df = self._process_input(filename, data, columns) temp_file = None ds_result = None @@ -107,7 +112,8 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic else: pipeline_api = AugerPipelineApi(self.ctx, None, model_id) predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, - predicted_at=predicted_at, no_features_in_result=no_features_in_result) + predicted_at=predicted_at, no_features_in_result=no_features_in_result, + score=score, score_true_data=score_true_data) try: ds_result = DataFrame.create_dataframe(predictions.get('signed_prediction_url'), @@ -137,7 +143,8 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic if temp_file: fsclient.remove_file(temp_file) - def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, + output, no_features_in_result, score, score_true_data): from auger_ml.model_exporter import ModelExporter is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) @@ -148,6 +155,11 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre if columns is not None: columns = list(columns) + if score and score_true_data is None: + options = fsclient.read_json_file(os.path.join(model_path, "options.json")) + ds = DataFrame.create_dataframe(filename_arg, data, [options['targetFeature']]) + score_true_data = ds.df + res, options = ModelExporter({}).predict_by_model_to_ds(model_path, path_to_predict=filename_arg, records=data, features=columns, threshold=threshold, no_features_in_result=no_features_in_result) @@ -158,16 +170,25 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre if isinstance(data, pd.DataFrame): ds_result.from_pandas = True - return ModelHelper.save_prediction(ds_result, + predictions = ModelHelper.save_prediction(ds_result, prediction_id = None, json_result=False, count_in_result=False, prediction_date=predicted_at, model_path=model_path, model_id=model_id, output=output) + if not score: + return predictions + + scores = ModelExporter({}).score_by_model(model_path, predictions=predictions, + test_path = score_true_data) + + return {'predicted': predictions, 'scores': scores} + # return ModelExporter({}).predict_by_model(model_path=model_path, # path_to_predict=filename_arg, records=data, features=columns, # threshold=threshold, prediction_date=predicted_at, # no_features_in_result=no_features_in_result) #, output=output) - def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, + output, no_features_in_result, score, score_true_data): model_deploy = ModelDeploy(self.ctx, None) is_model_loaded, model_path = model_deploy.verify_local_model(model_id, add_model_folder=False) if not is_model_loaded: @@ -180,7 +201,7 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co filename = os.path.join(self.ctx.config.get_path(), '.augerml', 'predict_data.csv') ds.saveToCsvFile(filename, compression=None) - predicted = self._docker_run_predict(filename, threshold, model_path) + predicted = self._docker_run_predict(filename, threshold, model_path, score, score_true_data) if not filename_arg: ds_result = DataFrame.create_dataframe(predicted) @@ -198,13 +219,14 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co return predicted - def _docker_run_predict(self, filename, threshold, model_path): + def _docker_run_predict(self, filename, threshold, model_path, score, score_true_data): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) docker_tag = cluster_settings.get('kubernetes_stack') predict_file = os.path.basename(filename) data_path = os.path.abspath(os.path.dirname(filename)) model_path = os.path.abspath(model_path) + #TODO: support score, score_true_data call_args = "--verbose=True --path_to_predict=./model_data/%s %s" % \ (predict_file, "--threshold=%s" % str(threshold) if threshold else '') diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 84f8c13a..d9011329 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -21,10 +21,12 @@ def deploy(self, project, model_id, locally, review, name, algorithm, score, dat @error_handler @authenticated - @with_project(autocreate=False) - def predict(self, project, filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result): - predicted = Model(self.ctx, project).predict( - filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + #@with_project(autocreate=False) + def predict(self, filename, model_id, threshold, locally, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data): + predicted = Model(self.ctx, project=None).predict( + filename, model_id, threshold, locally, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) if filename: self.ctx.log('Predictions stored in %s' % predicted) diff --git a/a2ml/api/utils/dataframe.py b/a2ml/api/utils/dataframe.py index caf7c45b..ca737052 100644 --- a/a2ml/api/utils/dataframe.py +++ b/a2ml/api/utils/dataframe.py @@ -50,6 +50,9 @@ def create_dataframe(data_path=None, records=None, features=None): elif records is not None and isinstance(records, pd.DataFrame): ds = DataFrame({}) ds.df = records + if features: + ds.df = ds.df[features] + ds.from_pandas = True else: ds = DataFrame({}) diff --git a/a2ml/cmdl/commands/cmd_deploy.py b/a2ml/cmdl/commands/cmd_deploy.py index cd050865..0920c958 100644 --- a/a2ml/cmdl/commands/cmd_deploy.py +++ b/a2ml/cmdl/commands/cmd_deploy.py @@ -4,10 +4,10 @@ @click.command('deploy', short_help='Deploy trained model.') [email protected]('model-id', required=False, type=click.STRING) @click.option('--provider', '-p', type=click.Choice(['auger','azure','external']), required=False, help='Cloud AutoML Provider.') [email protected]('model-id', required=False, type=click.STRING) [email protected]('--locally', is_flag=True, default=False, [email protected]('--locally', '-l', is_flag=True, default=False, help='Download and deploy trained model locally.') @click.option('--no-review', is_flag=True, default=False, help='Do not support model review based on actual data.') diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index 8fedc0a6..c9291fbc 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -28,15 +28,19 @@ def cmdl(ctx): @pass_context def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score, data_path): """Deploy trained model.""" - A2MLModel(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score, data_path=data_path) + A2MLModel(ctx, provider).deploy(model_id, locally=locally, review=not no_review, + name=name, algorithm=algorithm, score=score, data_path=data_path) @click.command('predict', short_help='Predict with deployed model.') [email protected]('model-id', required=True, type=click.STRING) @click.argument('filename', required=True, type=click.STRING) @click.option('--threshold', '-t', default=None, type=float, help='Threshold.') [email protected]('--model-id', '-m', type=click.STRING, required=True, - help='Deployed model id.') [email protected]('--locally', is_flag=True, default=False, [email protected]('--score', '-s', is_flag=True, default=False, + help='Calculate scores for predicted results.') [email protected]('--score_true_path', type=click.STRING, required=False, + help='Path to true values to calculate scores. If missed, target from filename used for true values.') [email protected]('--locally', '-l', is_flag=True, default=False, help='Predict locally using auger.ai.predict package.') @click.option('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @@ -50,12 +54,13 @@ def predict(ctx, provider, filename, model_id, threshold, locally, docker, outpu if docker: locally = "docker" - A2MLModel(ctx, provider).predict(filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) + A2MLModel(ctx, provider).predict(model_id, filename=filename, + threshold=threshold, score=score, score_true_data=score_true_path, + locally=locally, output=output) @click.command('actuals', short_help='Send actual values for deployed model. Needed for review and monitoring.') [email protected]('model-id', required=True, type=click.STRING) @click.argument('filename', required=True, type=click.STRING) [email protected]('--model-id', '-m', type=click.STRING, required=True, - help='Deployed model id.') @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') @click.option('--locally', is_flag=True, default=False, @@ -98,8 +103,7 @@ def undeploy(ctx, provider, model_id, locally): A2MLModel(ctx, provider).undeploy(model_id, locally) @click.command('delete_actuals', short_help='Delete files with actuals and predcitions locally or from specified provider(s).') [email protected]('--model-id', '-m', type=click.STRING, required=True, - help='Deployed model id.') [email protected]('model-id', required=True, type=click.STRING) @click.option('--with-predictions', is_flag=True, default=False, help='Remove predictions.') @click.option('--begin-date', '-b', type=click.STRING, required=False, diff --git a/a2ml/cmdl/commands/cmd_predict.py b/a2ml/cmdl/commands/cmd_predict.py index 526c31cb..b27c00aa 100644 --- a/a2ml/cmdl/commands/cmd_predict.py +++ b/a2ml/cmdl/commands/cmd_predict.py @@ -4,24 +4,28 @@ @click.command('predict', short_help='Predict with deployed model.') [email protected]('--provider', '-p', type=click.Choice(['auger','azure']), required=False, - help='Cloud AutoML Provider.') [email protected]('model-id', required=True, type=click.STRING) @click.argument('filename', required=True, type=click.STRING) @click.option('--threshold', '-t', default=None, type=float, help='Threshold.') [email protected]('--model-id', '-m', type=click.STRING, required=False, - help='Deployed model id.') [email protected]('--locally', is_flag=True, default=False, [email protected]('--score', '-s', is_flag=True, default=False, + help='Calculate scores for predicted results.') [email protected]('--score_true_path', type=click.STRING, required=False, + help='Path to true values to calculate scores. If missed, target from filename used for true values.') [email protected]('--locally', '-l', is_flag=True, default=False, help='Predict locally using auger.ai.predict package.') @click.option('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @click.option('--output', '-o', type=click.STRING, required=False, help='Output csv file path.') [email protected]('--provider', '-p', type=click.Choice(['auger','azure']), required=False, + help='Cloud AutoML Provider.') @pass_context def cmdl(ctx, provider, filename, model_id, threshold, locally, docker, output): """Predict with deployed model.""" ctx.setup_logger(format='') if docker: locally = "docker" - A2ML(ctx, provider).predict( - filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) + A2ML(ctx, provider).predict( model_id, filename=filename, + threshold=threshold, score=score, score_true_data=score_true_path, + locally=locally, output=output)	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-08-08T19:21:50	[]	[]
augerai/a2ml	augerai__a2ml-582	4963565d278901b20ce8715959270a36d4c2d053	diff --git a/a2ml/api/roi/var_names_fetcher.py b/a2ml/api/roi/var_names_fetcher.py index 2bbeb87c..355c8418 100644 --- a/a2ml/api/roi/var_names_fetcher.py +++ b/a2ml/api/roi/var_names_fetcher.py @@ -36,6 +36,9 @@ def evaluate_unary_op_node(self, node): def evaluate_func_node(self, node): list(map(lambda node: self.evaluate(node), node.arg_nodes)) + def evaluate_tuple_node(self, node): + list(map(lambda node: self.evaluate(node), node.item_nodes)) + def evaluate_top_node(self, node): list(map(lambda n: self.evaluate(n), node.child_nodes()))	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-07-05T18:24:00	[]	[]
augerai/a2ml	augerai__a2ml-578	2486bd45dfae8cf149e8dec88c18ba63b7f1141a	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index ac1a0a5d..cf42dc04 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.28' +__version__ = '1.0.29' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 32308b80..09fe4116 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -314,7 +314,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value. If missed - predict called automatically + * - predicted( or target): predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 9730d888..2be596f0 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -159,7 +159,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value. If missed - predict called automatically + * - predicted ( or target): predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index 46130f39..d641dc8c 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -169,8 +169,11 @@ def add_actuals( actuals_count = ds_actuals.count() ds_actuals.df.rename(columns={"actual": 'a2ml_actual'}, inplace=True) + if 'predicted' in ds_actuals.columns and not self.target_feature in ds_actuals.columns: + ds_actuals.df = ds_actuals.df.rename(columns={'predicted': self.target_feature}) + if provider is not None and (do_predict or not self.target_feature in ds_actuals.columns): - logging.info("Actual data missing predicted value column: %s. Call predict with features from actual data: %s"%(self.target_feature, ds_actuals.columns)) + logging.info("Actual data missing 'predicted' column and predicted value column: %s. Call predict with features from actual data: %s"%(self.target_feature, ds_actuals.columns)) self._do_predict(ctx, ds_actuals, provider) result = self._do_score_actual(ds_actuals.df)	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-07-02T19:47:17	[]	[]
augerai/a2ml	augerai__a2ml-560	6cf1f1c1e471b7101fbf8f510e8ef3cec6bacc88	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index dce8b34b..fa7c0d49 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.19' +__version__ = '1.0.20' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 0a727c3b..921f8892 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -221,7 +221,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, locally=False, provider=None): + threshold=None, output=None, no_features_in_result = None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -236,7 +236,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. output(str): Output csv file path. - locally(bool): Predicts using a local model if True, on the Provider Cloud if False. + no_features_in_result(bool) : Do not return feature columns in prediction result. False by default + locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. Returns: @@ -287,7 +288,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are returned as rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index da68a419..4df4de90 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -66,7 +66,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, locally=False, provider=None): + threshold=None, output=None, no_features_in_result=None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -81,7 +81,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. output(str): Output csv file path. - locally(bool): Predicts using a local model if True, on the Provider Cloud if False. + no_features_in_result(bool) : Do not return feature columns in prediction result. False by default + locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. Returns: @@ -132,7 +133,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are returned as rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 5bc25d38..3e998b5c 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -21,9 +21,9 @@ def evaluate(self, run_id = None): def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None ): return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) - def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): return AugerModel(self.ctx).predict( - model_id, filename, threshold, locally, data, columns, predicted_at, output) + model_id, filename, threshold, locally, data, columns, predicted_at, output, no_features_in_result) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): return AugerModel(self.ctx).actuals( diff --git a/a2ml/api/auger/impl/cloud/experiment_session.py b/a2ml/api/auger/impl/cloud/experiment_session.py index 7b6a474d..2668f8fe 100644 --- a/a2ml/api/auger/impl/cloud/experiment_session.py +++ b/a2ml/api/auger/impl/cloud/experiment_session.py @@ -72,6 +72,8 @@ def get_leaderboard(self): '{0:.4f}'.format(item.get('score_value')) } review_metric = self.ctx.config.get('review/metric') + if review_metric == 'roi': + review_metric = 'mda' if review_metric: l_item[review_metric] = \ '{0:.4f}'.format(item.get('raw_data', {}).get('all_scores', {}).get(review_metric, 0.0)) diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index 8149b184..e3232201 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -49,7 +49,7 @@ def check_endpoint(self, props=None): return is_endpoint - def predict(self, records, features, threshold=None, file_url=None, predicted_at=None): + def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') @@ -72,7 +72,9 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at prediction_api = AugerPredictionApi(self.ctx, self, use_endpoint=self.check_endpoint(props)) prediction_properties = \ - prediction_api.create(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at) + prediction_api.create(records, features, threshold=threshold, file_url=file_url, + predicted_at=predicted_at, no_features_in_result=no_features_in_result) + return prediction_properties.get('result') def actual(self, records, features, actuals_at, actuals_path, actual_date_column): diff --git a/a2ml/api/auger/impl/cloud/pipeline_file.py b/a2ml/api/auger/impl/cloud/pipeline_file.py index ddeee2ec..bb7e65db 100644 --- a/a2ml/api/auger/impl/cloud/pipeline_file.py +++ b/a2ml/api/auger/impl/cloud/pipeline_file.py @@ -31,7 +31,7 @@ def download(self, url, path_to_download, trial_id): return file_name def _get_status_name(self): - return 's3_model_path_status' + return 'signed_s3_model_path_status' def _log_status(self, status): if status is None: diff --git a/a2ml/api/auger/impl/cloud/prediction.py b/a2ml/api/auger/impl/cloud/prediction.py index 67f85dd5..c9c76704 100644 --- a/a2ml/api/auger/impl/cloud/prediction.py +++ b/a2ml/api/auger/impl/cloud/prediction.py @@ -14,7 +14,7 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self.parent_id_name = "endpoint_id" self._set_api_request_path("AugerEndpointPredictionApi") - def create(self, records, features, threshold=None, file_url=None, predicted_at=None): + def create(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): params = { 'records': records, 'features': features, @@ -33,4 +33,7 @@ def create(self, records, features, threshold=None, file_url=None, predicted_at= if predicted_at: params['predicted_at'] = str(predicted_at) + if no_features_in_result is not None: + params['no_features_in_result'] = no_features_in_result + return self._call_create(params, ['requested', 'running']) diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 3ece907e..c6afa0e7 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -30,23 +30,21 @@ def review(self, model_id): def undeploy(self, model_id, locally=False): return ModelUndeploy(self.ctx, self.project).execute(model_id, locally) - def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): if locally: self.deploy(model_id, locally) - return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output) + return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): if locally: - is_loaded, model_path, model_name = ModelDeploy(self.ctx, self.project).\ - verify_local_model(model_id) + is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) if not is_loaded: raise AugerException('Model should be deployed locally.') - model_path, model_existed = ModelPredict(self.ctx)._extract_model(model_name) params = { - 'model_path': os.path.join(model_path, "model"), + 'model_path': model_path, 'roi': { 'filter': str(self.ctx.config.get('review/roi/filter')), 'revenue': str(self.ctx.config.get('review/roi/revenue')), @@ -67,28 +65,22 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N def delete_actuals(self, model_id, with_predictions=False, begin_date=None, end_date=None, locally=False): if locally: - is_loaded, model_path, model_name = ModelDeploy(self.ctx, self.project).\ - verify_local_model(model_id) - + is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) if not is_loaded: raise AugerException('Model should be deployed locally.') - model_path, model_existed = ModelPredict(self.ctx)._extract_model(model_name) - return ModelReview({'model_path': os.path.join(model_path, "model")}).delete_actuals( + return ModelReview({'model_path': model_path}).delete_actuals( with_predictions=with_predictions, begin_date=begin_date, end_date=end_date) else: return ModelDeleteActual(self.ctx).execute(model_id, with_predictions, begin_date, end_date) def build_review_data(self, model_id, locally, output): if locally: - is_loaded, model_path, model_name = ModelDeploy(self.ctx, self.project).\ - verify_local_model(model_id) - + is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) if not is_loaded: raise AugerException('Model should be deployed locally.') - model_path, model_existed = ModelPredict(self.ctx)._extract_model(model_name) - return ModelReview({'model_path': os.path.join(model_path, "model")}).build_review_data( + return ModelReview({'model_path': model_path}).build_review_data( data_path=self.ctx.config.get("source"), output=output) else: raise Exception("Not Implemented.") diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index be68bfcc..9eeeaf2a 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -23,7 +23,7 @@ def __init__(self, ctx, project): def execute(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): if locally: - return self.deploy_model_locally(model_id, review, name, data_path) + return self.deploy_model_locally(model_id, review, name, data_path, locally) else: return self.deploy_model_in_cloud(model_id, review, name, algorithm, score, data_path) @@ -159,36 +159,65 @@ def deploy_model_in_cloud(self, model_id, review, name, algorithm, score, data_p return pipeline_properties.get('id') - def deploy_model_locally(self, model_id, review, name, data_path): - is_loaded, model_path, model_name = self.verify_local_model(model_id) + def deploy_model_locally(self, model_id, review, name, data_path, locally): + is_loaded, model_path = self.verify_local_model(model_id) #TODO: support review flag if not is_loaded: self.ctx.log('Downloading model %s' % model_id) self.project.start() + models_path = os.path.join(self.ctx.config.get_path(), 'models') pipeline_file_api = AugerPipelineFileApi(self.ctx, None) pipeline_file_properties = pipeline_file_api.create(model_id) downloaded_model_file = pipeline_file_api.download( pipeline_file_properties['signed_s3_model_path'], - model_path, model_id) + models_path, model_id) self.ctx.log('Downloaded model to %s' % downloaded_model_file) - self.ctx.log('Pulling docker image required to predict') - self._docker_pull_image() + if locally == 'docker': + self.ctx.log('Pulling docker image required to predict') + self._docker_pull_image() + else: + self.ctx.log('To run predict locally install a2ml[predict]') else: - self.ctx.log('Downloaded model is %s' % model_name) + self.ctx.log('Downloaded model is %s' % model_path) return model_id - def verify_local_model(self, model_id): - model_path = os.path.join(self.ctx.config.get_path(), 'models') - model_name = os.path.join(model_path, 'model-%s.zip' % model_id) - is_exists = fsclient.is_folder_exists(os.path.join(model_path,"model-%s"%model_id)) - if not is_exists: - is_exists = fsclient.is_file_exists(model_name) - return is_exists, model_path, model_name + def get_local_model_paths(self, model_id): + models_path = os.path.join(self.ctx.config.get_path(), 'models') + model_zip_path = os.path.join(models_path, 'model-%s.zip' % model_id) + model_path = os.path.join(models_path,"model-%s"%model_id) + + return model_path, model_zip_path + + def verify_local_model(self, model_id, add_model_folder=True): + model_path, model_zip_path = self.get_local_model_paths(model_id) + + is_exists = fsclient.is_folder_exists(model_path) + if not is_exists and fsclient.is_file_exists(model_zip_path): + self._extract_model(model_zip_path) + + if add_model_folder: + model_path = os.path.join(model_path, "model") + + is_exists = fsclient.is_folder_exists(model_path) + + return is_exists, model_path + + def _extract_model(self, model_name): + from zipfile import ZipFile + + model_path = os.path.splitext(model_name)[0] + model_existed = os.path.exists(model_path) + + if not model_existed: + with ZipFile(model_name, 'r') as zip_file: + zip_file.extractall(model_path) + + return model_path, model_existed def _docker_pull_image(self): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) diff --git a/a2ml/api/auger/impl/mparts/predict.py b/a2ml/api/auger/impl/mparts/predict.py index 510e4ce4..005de202 100644 --- a/a2ml/api/auger/impl/mparts/predict.py +++ b/a2ml/api/auger/impl/mparts/predict.py @@ -1,7 +1,6 @@ import os import shutil import subprocess -from zipfile import ZipFile import sys import pandas as pd @@ -23,16 +22,20 @@ def __init__(self, ctx): super(ModelPredict, self).__init__() self.ctx = ctx - def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, + predicted_at=None, output=None, no_features_in_result=None): if filename and not (filename.startswith("http:") or filename.startswith("https:")) and\ not fsclient.is_s3_path(filename): self.ctx.log('Predicting on data in %s' % filename) filename = os.path.abspath(filename) if locally: - predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output) + if locally == "docker": + predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + else: + predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) else: - predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output) + predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) return predicted @@ -95,7 +98,7 @@ def _check_model_project(self, pipeline_api): raise AugerException("Project name: %s in config.yml is different from model project name: %s. Please change name in config.yml."%( self.ctx.config.get('name'), model_project_name)) - def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, output): + def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): records, features, file_url, is_pandas_df = self._process_input(filename, data, columns) temp_file = None ds_result = None @@ -103,7 +106,8 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic ds_result = DataFrame.create_dataframe(None, [], features+[self.ctx.config.get('target')]) else: pipeline_api = AugerPipelineApi(self.ctx, None, model_id) - predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at) + predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, + predicted_at=predicted_at, no_features_in_result=no_features_in_result) try: ds_result = DataFrame.create_dataframe(predictions.get('signed_prediction_url'), @@ -121,11 +125,10 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic ds_result.loaded_columns = columns ds_result.from_pandas = is_pandas_df - is_model_loaded, model_path_1, model_name = \ - ModelDeploy(self.ctx, None).verify_local_model(model_id) - model_path = None - if is_model_loaded: - model_path = os.path.join(model_path_1, "model-%s"%model_id, 'model') + # Save prediction in local model folder if exist + is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) + if not is_model_loaded: + model_path = None return ModelHelper.save_prediction(ds_result, prediction_id = None, json_result=False, count_in_result=False, prediction_date=predicted_at, @@ -134,17 +137,42 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic if temp_file: fsclient.remove_file(temp_file) - def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, output): - model_deploy = ModelDeploy(self.ctx, None) - is_model_loaded, model_path, model_name = \ - model_deploy.verify_local_model(model_id) + def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + from auger_ml.model_exporter import ModelExporter + is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) if not is_model_loaded: raise AugerException('Model isn\'t loaded locally. ' 'Please use a2ml deploy command to download model.') - model_path, model_existed = self._extract_model(model_name) - model_options = fsclient.read_json_file(os.path.join(model_path, "model", "options.json")) + if columns is not None: + columns = list(columns) + + res, options = ModelExporter({}).predict_by_model_to_ds(model_path, + path_to_predict=filename_arg, records=data, features=columns, + threshold=threshold, no_features_in_result=no_features_in_result) + + ds_result = DataFrame({'data_path': None}) + ds_result.df = res.df + ds_result.loaded_columns = columns + if isinstance(data, pd.DataFrame): + ds_result.from_pandas = True + + return ModelHelper.save_prediction(ds_result, + prediction_id = None, json_result=False, count_in_result=False, prediction_date=predicted_at, + model_path=model_path, model_id=model_id, output=output) + + # return ModelExporter({}).predict_by_model(model_path=model_path, + # path_to_predict=filename_arg, records=data, features=columns, + # threshold=threshold, prediction_date=predicted_at, + # no_features_in_result=no_features_in_result) #, output=output) + + def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + model_deploy = ModelDeploy(self.ctx, None) + is_model_loaded, model_path = model_deploy.verify_local_model(model_id, add_model_folder=False) + if not is_model_loaded: + raise AugerException('Model isn\'t loaded locally. ' + 'Please use a2ml deploy command to download model.') filename = filename_arg if not filename: @@ -152,15 +180,7 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre filename = os.path.join(self.ctx.config.get_path(), '.augerml', 'predict_data.csv') ds.saveToCsvFile(filename, compression=None) - try: - predicted = \ - self._docker_run_predict(filename, threshold, model_path) - finally: - # clean up unzipped model - # if it wasn't unzipped before - if not model_existed: - fsclient.remove_folder(model_path) - model_path = None + predicted = self._docker_run_predict(filename, threshold, model_path) if not filename_arg: ds_result = DataFrame.create_dataframe(predicted) @@ -178,16 +198,6 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre return predicted - def _extract_model(self, model_name): - model_path = os.path.splitext(model_name)[0] - model_existed = os.path.exists(model_path) - - if not model_existed: - with ZipFile(model_name, 'r') as zip_file: - zip_file.extractall(model_path) - - return model_path, model_existed - def _docker_run_predict(self, filename, threshold, model_path): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) docker_tag = cluster_settings.get('kubernetes_stack') diff --git a/a2ml/api/auger/impl/mparts/undeploy.py b/a2ml/api/auger/impl/mparts/undeploy.py index 1dc26592..c15e2af2 100644 --- a/a2ml/api/auger/impl/mparts/undeploy.py +++ b/a2ml/api/auger/impl/mparts/undeploy.py @@ -17,15 +17,11 @@ def __init__(self, ctx, project): def execute(self, model_id, locally=False): if locally: - is_loaded, model_path, model_name = \ - ModelDeploy(self.ctx, self.project).verify_local_model(model_id) - self.ctx.log("Undeploy model. Remove local model: %s" % model_name) + model_path, model_zip_path = ModelDeploy(self.ctx, self.project).get_local_model_paths(model_id) + self.ctx.log("Undeploy model. Remove local model: %s" % model_path) - if is_loaded: - fsclient.remove_file(model_name) - - model_folder = os.path.splitext(model_name)[0] - fsclient.remove_folder(model_folder) + fsclient.remove_file(model_zip_path) + fsclient.remove_folder(model_path) else: pipeline_api = AugerPipelineApi(self.ctx, None, model_id) if pipeline_api.check_endpoint(): diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 24bbd661..84f8c13a 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -22,9 +22,9 @@ def deploy(self, project, model_id, locally, review, name, algorithm, score, dat @error_handler @authenticated @with_project(autocreate=False) - def predict(self, project, filename, model_id, threshold, locally, data, columns, predicted_at, output): + def predict(self, project, filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result): predicted = Model(self.ctx, project).predict( - filename, model_id, threshold, locally, data, columns, predicted_at, output) + filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) if filename: self.ctx.log('Predictions stored in %s' % predicted) diff --git a/a2ml/api/azure/a2ml.py b/a2ml/api/azure/a2ml.py index 078a7c9b..ef265de8 100644 --- a/a2ml/api/azure/a2ml.py +++ b/a2ml/api/azure/a2ml.py @@ -26,12 +26,12 @@ def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None return AzureModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) - def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): from a2ml.api.azure.model import AzureModel return AzureModel(self.ctx).predict( filename, model_id, threshold=threshold, locally=locally, data=data, columns=columns, - predicted_at=predicted_at, output=output) + predicted_at=predicted_at, output=output, no_features_in_result=no_features_in_result) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): from a2ml.api.azure.model import AzureModel diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index 5f7fb49f..8b473826 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -239,8 +239,8 @@ def get_df(data): @error_handler @authenticated def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, - predicted_at=None, output=None, json_result=False, count_in_result=False, prediction_id=None - ): + predicted_at=None, output=None, json_result=False, count_in_result=False, prediction_id=None, + no_features_in_result=None): ds = DataFrame.create_dataframe(filename, data, columns) model_path = self.ctx.config.get_model_path(model_id) options = fsclient.read_json_file(os.path.join(model_path, "options.json")) diff --git a/a2ml/api/model_review/model_helper.py b/a2ml/api/model_review/model_helper.py index 355b49a1..d2244c38 100644 --- a/a2ml/api/model_review/model_helper.py +++ b/a2ml/api/model_review/model_helper.py @@ -100,7 +100,7 @@ def save_metric(metric_id, project_path, metric_name, metric_data): # @staticmethod # def _get_score_byname(scoring): - # from sklearn.metrics.scorer import get_scorer + # from sklearn.metrics import get_scorer # from sklearn.metrics import SCORERS # #TODO: below metrics does not directly map to sklearn: @@ -157,7 +157,7 @@ def save_metric(metric_id, project_path, metric_name, metric_data): @staticmethod def calculate_scores(options, y_test, X_test=None, estimator=None, y_pred=None, raise_main_score=True): - from sklearn.metrics.scorer import get_scorer + from sklearn.metrics import get_scorer from sklearn.model_selection._validation import _score from sklearn.metrics import confusion_matrix diff --git a/a2ml/api/model_review/scores/classification.py b/a2ml/api/model_review/scores/classification.py index 76493cc3..3acd890d 100644 --- a/a2ml/api/model_review/scores/classification.py +++ b/a2ml/api/model_review/scores/classification.py @@ -3,7 +3,7 @@ from sklearn.metrics import make_scorer, recall_score, average_precision_score, roc_auc_score from sklearn.metrics import confusion_matrix from sklearn.metrics import matthews_corrcoef as mcc -from sklearn.metrics.scorer import SCORERS +from sklearn.metrics import SCORERS def kappa(y_true, y_pred, weights=None, allow_off_by_one=False): diff --git a/a2ml/api/model_review/scores/regression.py b/a2ml/api/model_review/scores/regression.py index b8cdf0bb..704f4539 100644 --- a/a2ml/api/model_review/scores/regression.py +++ b/a2ml/api/model_review/scores/regression.py @@ -1,6 +1,6 @@ import numpy as np from sklearn.metrics import make_scorer, mean_squared_error, mean_squared_log_error, mean_absolute_error -from sklearn.metrics.scorer import SCORERS +from sklearn.metrics import SCORERS EPSILON = 1e-10 diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index a380e6ac..8fedc0a6 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -37,14 +37,19 @@ def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score, @click.option('--model-id', '-m', type=click.STRING, required=True, help='Deployed model id.') @click.option('--locally', is_flag=True, default=False, + help='Predict locally using auger.ai.predict package.') [email protected]('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') @click.option('--output', '-o', type=click.STRING, required=False, help='Output csv file path.') @pass_context -def predict(ctx, provider, filename, model_id, threshold, locally, output): +def predict(ctx, provider, filename, model_id, threshold, locally, docker, output): """Predict with deployed model.""" + if docker: + locally = "docker" + A2MLModel(ctx, provider).predict(filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) @click.command('actuals', short_help='Send actual values for deployed model. Needed for review and monitoring.') diff --git a/a2ml/cmdl/commands/cmd_predict.py b/a2ml/cmdl/commands/cmd_predict.py index 5665273d..526c31cb 100644 --- a/a2ml/cmdl/commands/cmd_predict.py +++ b/a2ml/cmdl/commands/cmd_predict.py @@ -12,12 +12,16 @@ @click.option('--model-id', '-m', type=click.STRING, required=False, help='Deployed model id.') @click.option('--locally', is_flag=True, default=False, + help='Predict locally using auger.ai.predict package.') [email protected]('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @click.option('--output', '-o', type=click.STRING, required=False, help='Output csv file path.') @pass_context -def cmdl(ctx, provider, filename, model_id, threshold, locally, output): +def cmdl(ctx, provider, filename, model_id, threshold, locally, docker, output): """Predict with deployed model.""" ctx.setup_logger(format='') + if docker: + locally = "docker" A2ML(ctx, provider).predict( filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) diff --git a/a2ml/tasks_queue/tasks_hub_api.py b/a2ml/tasks_queue/tasks_hub_api.py index 534adde5..5340fca5 100644 --- a/a2ml/tasks_queue/tasks_hub_api.py +++ b/a2ml/tasks_queue/tasks_hub_api.py @@ -531,7 +531,8 @@ def predict_by_model_task(params): count_in_result=params.get('count_in_result'), predicted_at=params.get('prediction_date'), prediction_id = params.get('prediction_id'), - locally = params.get('locally', False) + locally = params.get('locally', False), + no_features_in_result=params.get('no_features_in_result', False) ) _update_hub_objects(ctx, params.get('provider'), params) diff --git a/setup.py b/setup.py index cf55b2a8..1a73562d 100644 --- a/setup.py +++ b/setup.py @@ -28,9 +28,9 @@ def run(self): install_requires = [ - 'numpy<1.19.0,>=1.16.0', # version for azure - 'pandas>=0.22', # version for azure - 'joblib>=0.14.1', # version for azure + 'numpy==1.18.5', # version for azure + 'pandas==1.2.4', # version for azure + 'joblib==1.0.1', # version for azure 'ruamel.yaml>0.16.7', # version for azure 'pyarrow<2.0.0,>=0.17.0', # version for azure 'scipy==1.5.2', @@ -72,23 +72,30 @@ def run(self): 'redis', 's3fs>=0.4.0,<0.5.0', 'uvicorn', + 'scikit-learn==0.24.2' ], 'azure': [ - 'scikit-learn~=0.22.2', - 'xgboost<=0.90', + #'scikit-learn~=0.22.2', + #'xgboost<=0.90', # https://github.com/Azure/azure-sdk-for-python/issues/13871 #'azure-mgmt-resource==10.2.0', - 'azureml-sdk[automl]~=1.22.0' + #this needs to move to setup.azure.py and do not include default + 'azureml-sdk[automl]==1.29.0' ], 'google': [ 'google-cloud-automl' + ], + 'predict': [ + 'auger.ai.predict==1.0.72' ] } # Meta dependency groups. all_deps = [] for group_name in extras: - all_deps += extras[group_name] + if group_name != 'predict' and group_name != 'google' and group_name != 'azure': + all_deps += extras[group_name] + extras['all'] = all_deps	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-05-27T20:31:05	[]	[]
augerai/a2ml	augerai__a2ml-547	84094892b7a1865b4d6a0f1d13b2af695b5f154b	diff --git a/a2ml/api/roi/grammar.bnf b/a2ml/api/roi/grammar.bnf index 73b56ba7..5701ae44 100644 --- a/a2ml/api/roi/grammar.bnf +++ b/a2ml/api/roi/grammar.bnf @@ -96,7 +96,7 @@ func_call_statement: \| NAME '(' (expression (',' expression)*)? ')' top_expression: - \| top_expression_type' NUMBER 'by' expression ['per' expression] ['where' expression] ['from' '(' top_expression ')'] + \| top_expression_type' NUMBER 'by' shift_expr ['per' expression] ['where' expression] ['from' '(' top_expression ')'] top_expression_type: \| 'top' diff --git a/a2ml/api/roi/parser.py b/a2ml/api/roi/parser.py index d2b798fa..d9c2a920 100644 --- a/a2ml/api/roi/parser.py +++ b/a2ml/api/roi/parser.py @@ -376,7 +376,7 @@ def top_expression(self): node.limit_node = self.const_node(Token.INT_CONST) self.eat(Token.BY) - node.order_node = self.expression() + node.order_node = self.shift_expr() if self.current_token.type == Token.PER: self.eat(Token.PER) diff --git a/a2ml/api/roi/validator.py b/a2ml/api/roi/validator.py index 03c2de16..b299bb03 100644 --- a/a2ml/api/roi/validator.py +++ b/a2ml/api/roi/validator.py @@ -1,6 +1,6 @@ from a2ml.api.roi.base_interpreter import BaseInterpreter from a2ml.api.roi.lexer import AstError, Lexer -from a2ml.api.roi.parser import Parser +from a2ml.api.roi.parser import Parser, TopNode class ValidationError(AstError): pass @@ -84,5 +84,8 @@ def evaluate_func_node(self, node): raise ValidationError(f"unknown function '{node.func_name}' at position {node.position()}") def evaluate_top_node(self, node): + if not isinstance(self.root, TopNode): + raise ValidationError(f"top or bottom expression cannot be used as an argument or operand") + return all(map(lambda n: self.evaluate(n), node.child_nodes()))	Check how and/or works in top expression e.g. ``` top 8 by P and $spread_pct < 0.5 from (top 1 by P per $symbol) $spread_pct < 0.5 and top 8 by P from (top 1 by P per $symbol) ```	2021-04-26T13:57:29	[]	[]
augerai/a2ml	augerai__a2ml-545	0369a5aedaf72a0533c6e3890c1d668a8246fc23	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index cc08f086..9b076950 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.14' +__version__ = '1.0.15' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index aa213b05..0a727c3b 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -300,7 +300,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value + * - target: predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa @@ -310,7 +310,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N - Iris-virginica - Iris-virginica - It may also contain train features to retrain while Review(if target missed) and for distribution chart + It may also contain train features to predict(if target missed), retrain model while Review and for distribution chart This method support only one provider diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index c8764b19..da68a419 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -145,7 +145,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value + * - target: predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa @@ -155,7 +155,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N - Iris-virginica - Iris-virginica - It may also contain train features to retrain while Review(if target missed) and for distribution chart + It may also contain train features to predict(if target missed), retrain model while Review and for distribution chart This method support only one provider @@ -226,6 +226,12 @@ def review_alert(self, model_id, parameters = None, locally=False, provider=None * threshold (float) * sensitivity (int): The amount of time(in hours) this metric must be at or below the threshold to trigger the alert. + * threshold_policy (all_values/average_value/any_value) + + - all_values: Default value. Trigger an alert when all values in sensitivity below threshold + - average_value: Trigger an alert when average of values in sensitivity below threshold + - any_value: Trigger an alert when any value in sensitivity below threshold + * action (no/retrain/retrain_deploy) - no: no action should be executed diff --git a/a2ml/api/auger/impl/cloud/endpoint.py b/a2ml/api/auger/impl/cloud/endpoint.py index 27030333..e581af4a 100644 --- a/a2ml/api/auger/impl/cloud/endpoint.py +++ b/a2ml/api/auger/impl/cloud/endpoint.py @@ -14,8 +14,9 @@ def __init__(self, ctx, endpoint_api, endpoint_id=None): def create(self, pipeline_id, name): return self._call_create({'pipeline_id': pipeline_id, 'name': name},[]) - def update(self, name): - return self._call_update({ 'id': self.object_id, 'name': name}) + def update(self, params): + params['id'] = self.object_id + return self._call_update(params) def update_roi(self): roi_names = ['review/roi/filter', 'review/roi/investment', 'review/roi/revenue'] diff --git a/a2ml/api/auger/impl/cloud/experiment.py b/a2ml/api/auger/impl/cloud/experiment.py index 65194f98..45a6c715 100644 --- a/a2ml/api/auger/impl/cloud/experiment.py +++ b/a2ml/api/auger/impl/cloud/experiment.py @@ -96,8 +96,6 @@ def get_experiment_options(config, ): if config.get('experiment/search_space', None) is not None: options['search_space'] = config.get('experiment/search_space') - if config.get('review/metric'): - options['review_metric'] = config.get('review/metric') if config.get('review/alert/retrain_policy/type'): options['retrain_policy_type'] = config.get('review/alert/retrain_policy/type') if config.get('review/alert/retrain_policy/value'): diff --git a/a2ml/api/auger/impl/cloud/review_alert.py b/a2ml/api/auger/impl/cloud/review_alert.py index 30363621..821279ed 100644 --- a/a2ml/api/auger/impl/cloud/review_alert.py +++ b/a2ml/api/auger/impl/cloud/review_alert.py @@ -23,6 +23,7 @@ def create_update(self, parameters=None): 'active': parameters.get('active', config.get('review/alert/active')), 'kind': parameters.get('type', config.get('review/alert/type')), 'threshold': float(parameters.get('threshold', config.get('review/alert/threshold'))), + 'threshold_policy': parameters.get('threshold_policy', config.get('review/alert/threshold_policy')), 'sensitivity': int(parameters.get('sensitivity', config.get('review/alert/sensitivity'))), 'actions': parameters.get('action', config.get('review/alert/action')), 'notifications': parameters.get('notification', config.get('review/alert/notification')) diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index 61a32901..ddb7c98a 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -50,8 +50,12 @@ def create_update_review_alert(self, model_id, pipeline_properties=None, paramet endpoint_api = AugerEndpointApi(self.ctx, None, pipeline_properties['endpoint_pipelines'][0].get('endpoint_id')) + params = {'review_metric': self.ctx.config.get('review/metric')} if name and update_name: - endpoint_api.update(name) + params['name'] = name + + if params: + endpoint_api.update(params) session_id = endpoint_api.properties().get('primary_experiment_session_id') if session_id: diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index 43eae7b5..5f7fb49f 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -47,7 +47,6 @@ def deploy(self, model_id, locally, review, name=None, algorithm=None, score=Non 'scoreNames': [self.ctx.config.get('experiment/metric')], 'scoring': self.ctx.config.get('experiment/metric'), "score_name": self.ctx.config.get('experiment/metric'), - "review_metric": self.ctx.config.get('review/metric'), "originalFeatureColumns": model_features, "model_type": self.ctx.config.get("model_type") } diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index eec8cf41..916659c1 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -332,7 +332,6 @@ def score_model_performance_daily(self, date_from, date_to, extra_features=[]): res[str(curr_date)] = { 'scores': scores, 'score_name': self.options.get('score_name'), - 'review_metric': self.options.get('review_metric'), 'baseline_scores': baseline_score } diff --git a/a2ml/cmdl/template/config.yaml b/a2ml/cmdl/template/config.yaml index 566e9a90..2bfc8cc3 100644 --- a/a2ml/cmdl/template/config.yaml +++ b/a2ml/cmdl/template/config.yaml @@ -72,13 +72,19 @@ review: alert: # Activate/Deactivate Review Alert active: True - #model_accuracy - Decrease in Model Accuracy: the model accuracy threshold allowed before trigger is initiated. Default threshold: 0.7. Default sensitivity: 72 - #feature_average_range - Feature Average Out-Of-Range: Trigger an alert if average feature value during time period goes beyond the standard deviation range calculated during training period by the specified number of times or more. Default threshold: 1. Default sensitivity: 168 - #runtime_errors_burst - Burst Of Runtime Errors: Trigger an alert if runtime error count exceeds threshold. Default threshold: 5. Default sensitivity: 1 + # model_accuracy - Decrease in Model Accuracy: the model accuracy threshold allowed before trigger is initiated. Default threshold: 0.7. Default sensitivity: 72 + # feature_average_range - Feature Average Out-Of-Range: Trigger an alert if average feature value during time period goes beyond the standard deviation range calculated during training period by the specified number of times or more. Default threshold: 1. Default sensitivity: 168 + # runtime_errors_burst - Burst Of Runtime Errors: Trigger an alert if runtime error count exceeds threshold. Default threshold: 5. Default sensitivity: 1 type: model_accuracy threshold: 0.7 #The amount of time(in hours) this metric must be at or below the threshold to trigger the alert. sensitivity: 72 + + # all_values - Trigger an alert when all values in sensitivity below threshold. + # average_value - Trigger an alert when average of values in sensitivity below threshold. + # any_value - Trigger an alert when any value in sensitivity below threshold + threshold_policy: all_values + #no - no action should be executed #retrain - Use new predictions and actuals as test set to retrain the model. #retrain_deploy - Deploy retrained model and make it active model of this endpoint. diff --git a/docs/source/dev/configuration.rst b/docs/source/dev/configuration.rst index a8d2caee..c983ac6f 100644 --- a/docs/source/dev/configuration.rst +++ b/docs/source/dev/configuration.rst @@ -44,6 +44,7 @@ All Providers type: model_accuracy threshold: 0.7 sensitivity: 72 + threshold_policy: all_values action: retrain_deploy notification: user @@ -91,6 +92,13 @@ All Providers * review.alert.threshold Float * review.alert.sensitivity The amount of time(in hours) this metric must be at or below the threshold to trigger the alert. + * review.alert.threshold_policy + + * Supported Review Alert threshold policies + * all_values Trigger an alert when all values in sensitivity below threshold. + * average_value Trigger an alert when average of values in sensitivity below threshold. + * any_value Trigger an alert when any value in sensitivity below threshold. + * review.alert.action * Supported Review Alert actions diff --git a/docs/source/dev/mlram.rst b/docs/source/dev/mlram.rst index 941cf7d6..1da9d1cf 100644 --- a/docs/source/dev/mlram.rst +++ b/docs/source/dev/mlram.rst @@ -11,6 +11,8 @@ Auger model Monitored model =================== +See application example: https://github.com/augerai/mlram_apps + 1. Create A2ML application with external provider: .. code-block:: bash @@ -66,6 +68,9 @@ To review distribution chart , send training features with target and actuals: .. code-block:: python + # If call just after actuals, wait some time till server process the data + time.sleep(30) + ctx = Context() result = A2ML(ctx).review(model_id='external_model_id') if result['data']['status'] == 'retrain':	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-04-21T11:17:28	[]	[]
augerai/a2ml	augerai__a2ml-536	71d7180c37deba8a27bee7b1c67f9dae2d39553e	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index b49950b1..0fb5dff1 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.11' +__version__ = '1.0.12' diff --git a/a2ml/api/auger/impl/cloud/base.py b/a2ml/api/auger/impl/cloud/base.py index 81c45ac1..5ea06a53 100644 --- a/a2ml/api/auger/impl/cloud/base.py +++ b/a2ml/api/auger/impl/cloud/base.py @@ -134,6 +134,7 @@ def _call_create(self, params=None, progress=None,has_return_object=True): if has_return_object: if object_properties: self.object_id = object_properties.get('id') + self.object_name = object_properties.get('name') #name can be changed by hub if progress: self.wait_for_status(progress) return self.properties()	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-03-22T06:21:43	[]	[]
augerai/a2ml	augerai__a2ml-535	3088a5db6d2dc5fdb4c4489740411cf140b77a62	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index d521168a..b49950b1 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.10' +__version__ = '1.0.11' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index c92955a7..3a14194f 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -175,7 +175,8 @@ def evaluate(self, run_id = None): return self.runner.execute('evaluate', run_id = run_id) @show_result - def deploy(self, model_id, locally=False, review=True, provider=None, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally=False, review=True, provider=None, + name=None, algorithm=None, score=None, data_path=None): """Deploy a model locally or to specified provider(s). Note: @@ -190,6 +191,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, name (str): Friendly name for the model. Used as name for Review Endpoint algorithm (str): Self-hosted model(external provider) algorithm name. score (float): Self-hosted model(external provider) score. + data_path (str): Data path to fit model when deploy. Return new deployed model-id Returns: :: @@ -214,7 +216,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, model_id = result['model_id'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', model_id, locally, review, name, algorithm, score) + return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', + model_id, locally, review, name, algorithm, score, data_path) @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 83db12dd..b5400042 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -25,7 +25,8 @@ def __init__(self, ctx, provider=None): self.local_runner = lambda: self.build_runner(ctx, provider, force_local=True) @show_result - def deploy(self, model_id, locally=False, review=True, provider=None, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally=False, review=True, provider=None, + name=None, algorithm=None, score=None, data_path=None): """Deploy a model locally or to specified provider(s). Args: @@ -36,6 +37,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, name (str): Friendly name for the model. Used as name for Review Endpoint algorithm (str): Self-hosted model(external provider) algorithm name. score (float): Self-hosted model(external provider) score. + data_path (str): Data path to fit model when deploy. Return new deployed model-id Returns: :: @@ -59,7 +61,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, model_id = result['model_id'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', model_id, locally, review, name, algorithm, score) + return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', + model_id, locally, review, name, algorithm, score, data_path) @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 015fff4e..5bc25d38 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -18,8 +18,8 @@ def train(self): def evaluate(self, run_id = None): return AugerExperiment(self.ctx).leaderboard(run_id) - def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): - return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score) + def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None ): + return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): return AugerModel(self.ctx).predict( diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index 7e289e20..8149b184 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -11,8 +11,9 @@ def __init__(self, ctx, experiment_api, pipeline_id=None): super(AugerPipelineApi, self).__init__( ctx, experiment_api, None, pipeline_id) - def create(self, trial_id, review=True, name=None): - return self._call_create({'trial_id': trial_id, 'is_review_model_enabled' : review, 'name': name}, + def create(self, trial_id, review=True, name=None, refit_data_path=None): + return self._call_create({'trial_id': trial_id, 'is_review_model_enabled' : review, 'name': name, + 'refit_data_path': refit_data_path}, ['creating_files', 'packaging', 'deploying']) def create_external(self, review, name, project_id, algorithm, score): diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 19ec06c7..3ece907e 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -18,8 +18,8 @@ def __init__(self, ctx, project): self.project = project self.ctx = ctx - def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): - return ModelDeploy(self.ctx, self.project).execute(model_id, locally, review, name, algorithm, score) + def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): + return ModelDeploy(self.ctx, self.project).execute(model_id, locally, review, name, algorithm, score, data_path) def review_alert(self, model_id, parameters, name): return ModelDeploy(self.ctx, self.project).create_update_review_alert(model_id, None, parameters, name) diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index 391cf465..90d020eb 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -21,11 +21,11 @@ def __init__(self, ctx, project): self.project = project self.ctx = ctx - def execute(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): + def execute(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): if locally: - return self.deploy_model_locally(model_id, review, name) + return self.deploy_model_locally(model_id, review, name, data_path) else: - return self.deploy_model_in_cloud(model_id, review, name, algorithm, score) + return self.deploy_model_in_cloud(model_id, review, name, algorithm, score, data_path) def create_update_review_alert(self, model_id, pipeline_properties=None, parameters=None, name=None): if not self.ctx.config.get('review'): @@ -121,7 +121,9 @@ def review(self, model_id): } return result - def deploy_model_in_cloud(self, model_id, review, name, algorithm, score): + def deploy_model_in_cloud(self, model_id, review, name, algorithm, score, data_path): + from .predict import ModelPredict + self.ctx.log('Deploying model %s' % model_id) if self.ctx.is_external_provider(): @@ -129,8 +131,12 @@ def deploy_model_in_cloud(self, model_id, review, name, algorithm, score): self.ctx, None).create_external(review, name, self.project.object_id, algorithm, score) else: self.project.start() + data_url = None + if data_path: + _, _, data_url, _ = ModelPredict(self.ctx)._process_input(data_path, None, None) + pipeline_properties = AugerPipelineApi( - self.ctx, None).create(model_id, review, name) + self.ctx, None).create(model_id, review, name, data_url) if pipeline_properties.get('status') == 'ready': if review: @@ -144,7 +150,7 @@ def deploy_model_in_cloud(self, model_id, review, name, algorithm, score): return pipeline_properties.get('id') - def deploy_model_locally(self, model_id, review, name): + def deploy_model_locally(self, model_id, review, name, data_path): is_loaded, model_path, model_name = self.verify_local_model(model_id) #TODO: support review flag if not is_loaded: diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index b901ee1c..24bbd661 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -15,8 +15,8 @@ def __init__(self, ctx): @error_handler @authenticated @with_project(autocreate=False) - def deploy(self, project, model_id, locally, review, name, algorithm, score): - model_id = Model(self.ctx, project).deploy(model_id, locally, review, name, algorithm, score) + def deploy(self, project, model_id, locally, review, name, algorithm, score, data_path): + model_id = Model(self.ctx, project).deploy(model_id, locally, review, name, algorithm, score, data_path) return {'model_id': model_id} @error_handler diff --git a/a2ml/api/azure/a2ml.py b/a2ml/api/azure/a2ml.py index a2647eee..078a7c9b 100644 --- a/a2ml/api/azure/a2ml.py +++ b/a2ml/api/azure/a2ml.py @@ -21,10 +21,10 @@ def evaluate(self, run_id = None): return AzureExperiment(self.ctx).leaderboard(run_id) - def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): from a2ml.api.azure.model import AzureModel - return AzureModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score) + return AzureModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): from a2ml.api.azure.model import AzureModel diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index 83195e43..43eae7b5 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -20,7 +20,7 @@ def __init__(self, ctx): @error_handler @authenticated - def deploy(self, model_id, locally, review, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally, review, name=None, algorithm=None, score=None, data_path=None): if locally: is_loaded, model_path = self.verify_local_model(model_id) if is_loaded: diff --git a/a2ml/cmdl/commands/cmd_deploy.py b/a2ml/cmdl/commands/cmd_deploy.py index 474b1dae..0793b065 100644 --- a/a2ml/cmdl/commands/cmd_deploy.py +++ b/a2ml/cmdl/commands/cmd_deploy.py @@ -17,8 +17,10 @@ help='Self-hosted model(external provider) algorithm name.') @click.option('--score', '-s', required=False, type=float, help='Self-hosted model(external provider) score.') [email protected]('--data-path', '-d', type=click.STRING, required=False, + help='Data path to fit model when deploy. Return new deployed model-id') @pass_context -def cmdl(ctx, provider, model_id, locally, no_review, name, algorithm, score): +def cmdl(ctx, provider, model_id, locally, no_review, name, algorithm, score, data_path): """Deploy trained model.""" ctx.setup_logger(format='') - A2ML(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score) + A2ML(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score, data_path=data_path) diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index ccc5bacf..e6b187b4 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -23,10 +23,12 @@ def cmdl(ctx): help='Self-hosted model(external provider) algorithm name.') @click.option('--score', '-s', required=False, type=float, help='Self-hosted model(external provider) score.') [email protected]('--data-path', '-d', type=click.STRING, required=False, + help='Data path to fit model when deploy. Return new deployed model-id') @pass_context -def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score): +def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score, data_path): """Deploy trained model.""" - A2MLModel(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score) + A2MLModel(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score, data_path=data_path) @click.command('predict', short_help='Predict with deployed model.') @click.argument('filename', required=True, type=click.STRING)	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-03-09T17:48:33	[]	[]
augerai/a2ml	augerai__a2ml-534	2e580917f270c6ed64528af1e34c08626d5a3fc0	diff --git a/a2ml/__init__.py b/a2ml/__init__.py index a67a991b..d521168a 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.09' +__version__ = '1.0.10' diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index ed58746d..83db12dd 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -204,7 +204,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', model_id, filename, data, columns, actuals_at, actual_date_column, locally) @show_result - def review_alert(self, model_id, parameters = None, locally=False, provider=None): + def review_alert(self, model_id, parameters = None, locally=False, provider=None, name=None): """Update Review parameters. Args: @@ -230,6 +230,7 @@ def review_alert(self, model_id, parameters = None, locally=False, provider=None locally(bool): Process review locally. provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider defined by model_id or set in costructor. + name (str): Friendly name for the model. Used as name for Review Endpoint Returns: :: @@ -244,7 +245,7 @@ def review_alert(self, model_id, parameters = None, locally=False, provider=None ctx = Context() model = A2MLModel(ctx).review_alert(model_id='D881079E1ED14FB') """ - return self.get_runner(locally, model_id, provider).execute_one_provider('review_alert', model_id, parameters) + return self.get_runner(locally, model_id, provider).execute_one_provider('review_alert', model_id, parameters, name) @show_result def review(self, model_id, locally=False, provider=None): diff --git a/a2ml/api/auger/impl/cloud/endpoint.py b/a2ml/api/auger/impl/cloud/endpoint.py index eed9eb81..27030333 100644 --- a/a2ml/api/auger/impl/cloud/endpoint.py +++ b/a2ml/api/auger/impl/cloud/endpoint.py @@ -14,6 +14,9 @@ def __init__(self, ctx, endpoint_api, endpoint_id=None): def create(self, pipeline_id, name): return self._call_create({'pipeline_id': pipeline_id, 'name': name},[]) + def update(self, name): + return self._call_update({ 'id': self.object_id, 'name': name}) + def update_roi(self): roi_names = ['review/roi/filter', 'review/roi/investment', 'review/roi/revenue'] roi_values = [] diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index aedc93d4..19ec06c7 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -21,8 +21,8 @@ def __init__(self, ctx, project): def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): return ModelDeploy(self.ctx, self.project).execute(model_id, locally, review, name, algorithm, score) - def review_alert(self, model_id, parameters): - return ModelDeploy(self.ctx, self.project).create_update_review_alert(model_id, None, parameters) + def review_alert(self, model_id, parameters, name): + return ModelDeploy(self.ctx, self.project).create_update_review_alert(model_id, None, parameters, name) def review(self, model_id): return ModelDeploy(self.ctx, self.project).review(model_id) diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index 1023a93b..391cf465 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -34,7 +34,8 @@ def create_update_review_alert(self, model_id, pipeline_properties=None, paramet if not pipeline_properties: pipeline_properties = AugerPipelineApi(self.ctx, None, model_id).properties() - endpoint_api = None + endpoint_api = None + update_name = True if not pipeline_properties.get('endpoint_pipelines'): self.ctx.log('Creating review endpoint ...') endpoint_api = AugerEndpointApi(self.ctx, None) @@ -42,12 +43,16 @@ def create_update_review_alert(self, model_id, pipeline_properties=None, paramet name = fsclient.get_path_base_name(self.ctx.config.get('source')) endpoint_properties = endpoint_api.create(pipeline_properties.get('id'), name) pipeline_properties['endpoint_pipelines'] = [endpoint_properties.get('id')] + update_name = False if pipeline_properties.get('endpoint_pipelines'): if endpoint_api is None: endpoint_api = AugerEndpointApi(self.ctx, None, pipeline_properties['endpoint_pipelines'][0].get('endpoint_id')) + if name and update_name: + endpoint_api.update(name) + session_id = endpoint_api.properties().get('primary_experiment_session_id') if session_id: AugerExperimentSessionApi(self.ctx, None, None, session_id).update_settings() diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 516c5c0f..b901ee1c 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -46,8 +46,8 @@ def delete_actuals(self, project, model_id, with_predictions=False, begin_date=N @error_handler @authenticated @with_project(autocreate=False) - def review_alert(self, project, model_id, parameters): - return Model(self.ctx, project).review_alert(model_id, parameters) + def review_alert(self, project, model_id, parameters, name): + return Model(self.ctx, project).review_alert(model_id, parameters, name) @error_handler @authenticated diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index e4342fd9..83195e43 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -554,7 +554,7 @@ def undeploy(self, model_id, locally): @error_handler @authenticated - def review_alert(self, model_id, parameters): + def review_alert(self, model_id, parameters, name): raise AzureException("Not Implemented. Set use_auger_cloud: True in config.yml") @error_handler diff --git a/a2ml/api/model_review/model_helper.py b/a2ml/api/model_review/model_helper.py index 64ceeb18..6a0c3541 100644 --- a/a2ml/api/model_review/model_helper.py +++ b/a2ml/api/model_review/model_helper.py @@ -4,7 +4,7 @@ import numpy as np import json -from a2ml.api.utils import get_uid, get_uid4, fsclient, remove_dups_from_list +from a2ml.api.utils import get_uid, get_uid4, fsclient, remove_dups_from_list, sort_arrays from a2ml.api.utils.dataframe import DataFrame @@ -184,6 +184,12 @@ def calculate_scores(options, y_test, X_test=None, estimator=None, y_pred=None, else: logging.error("calculate_scores: no scaling found for target fold group: %s"%options['fold_group']) + if options.get("score_top_count"): + if y_pred is None: + y_pred = estimator.predict(X_test) + + y_pred, y_test = sort_arrays(y_pred, y_test, options.get("score_top_count")) + all_scores = {} if y_pred is not None: if options.get('binaryClassification'): diff --git a/a2ml/api/utils/__init__.py b/a2ml/api/utils/__init__.py index 876bbf2e..f39f229e 100644 --- a/a2ml/api/utils/__init__.py +++ b/a2ml/api/utils/__init__.py @@ -249,4 +249,13 @@ def retry_helper(func, retry_errors=[], num_try=10, delay=10, ctx=None): time.sleep(delay*nTry) else: raise - + +def sort_arrays(ar1, ar2, top_n=None, desc=True): + p = ar1.argsort() + if desc: + p = p[::-1] + if top_n: + p = p[:top_n] + + return ar1[p], ar2[p] + diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index d2275b79..ccc5bacf 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -62,10 +62,12 @@ def actuals(ctx, provider, filename, model_id, locally): @click.argument('model-id', required=True, type=click.STRING) @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') [email protected]('--name', '-n', required=False, type=click.STRING, + help='Model friendly name.Used as name for Review Endpoint') @pass_context -def review_alert(ctx, provider, model_id): +def review_alert(ctx, provider, model_id, name): """Predict with deployed model.""" - A2MLModel(ctx, provider).review_alert(model_id) + A2MLModel(ctx, provider).review_alert(model_id, name=name) @click.command('review', short_help='Review information about deployed model.') @click.argument('model-id', required=True, type=click.STRING)	WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo	2021-03-03T11:26:13	[]	[]

augerai/a2ml

augerai__a2ml-611

30601dc95093e45ebafb500fa70fcdafe65edd24

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index e71bfca8..5baef1cb 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.79' +__version__ = '1.0.80' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 03454e01..b7002dab 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -279,7 +279,7 @@ def deploy(self, model_id, locally=False, review=False, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, threshold=None, score=False, score_true_data=None, - output=None, no_features_in_result = None, locally=False, provider=None): + output=None, no_features_in_result = None, locally=False, provider=None, predict_labels=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -299,6 +299,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. + predict_labels (dict, bool): Run ActiveLearn to select data for labelling Returns: if filename is not None. :: @@ -363,7 +364,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at """ return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result, - score, score_true_data ) + score, score_true_data, predict_labels ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 7f810fd8..770c9376 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -68,7 +68,7 @@ def deploy(self, model_id, locally=False, review=False, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, threshold=None, score=False, score_true_data=None, - output=None, no_features_in_result=None, locally=False, provider=None): + output=None, no_features_in_result=None, locally=False, provider=None, predict_labels=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -88,6 +88,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. + predict_labels (dict, bool): Run ActiveLearn to select data for labelling Returns: if filename is not None. :: @@ -152,7 +153,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at """ return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result, - score, score_true_data ) + score, score_true_data, predict_labels ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index f367c7a6..af40e9f8 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -25,10 +25,11 @@ def deploy(self, model_id, locally=False, review=False, name=None, algorithm=Non return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path, metadata) def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, - predicted_at=None, output=None, no_features_in_result=None, score=False, score_true_data=None): + predicted_at=None, output=None, no_features_in_result=None, score=False, + score_true_data=None, predict_labels=None): return AugerModel(self.ctx).predict( model_id, filename, threshold, locally, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, experiment_params=None, locally=False): diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index dc056474..aacf1a0e 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -95,7 +95,7 @@ def has_endpoint(self, props=None): return is_endpoint def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, - no_features_in_result=None, score=False, score_true_data=None): + no_features_in_result=None, score=False, score_true_data=None, predict_labels=None): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') @@ -120,7 +120,7 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at prediction_properties = \ prediction_api.create(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at, no_features_in_result=no_features_in_result, - score=score, score_true_data=score_true_data) + score=score, score_true_data=score_true_data, predict_labels=predict_labels) return prediction_properties.get('result') diff --git a/a2ml/api/auger/impl/cloud/prediction.py b/a2ml/api/auger/impl/cloud/prediction.py index 81f50387..da3dd44b 100644 --- a/a2ml/api/auger/impl/cloud/prediction.py +++ b/a2ml/api/auger/impl/cloud/prediction.py @@ -15,7 +15,7 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self._set_api_request_path("AugerEndpointPredictionApi") def create(self, records, features, threshold=None, file_url=None, predicted_at=None, - no_features_in_result=None, score=False, score_true_data=None): + no_features_in_result=None, score=False, score_true_data=None, predict_labels=None): params = { 'records': records, 'features': features, @@ -43,4 +43,7 @@ def create(self, records, features, threshold=None, file_url=None, predicted_at= if score_true_data: params['score_true_data'] = score_true_data + if predict_labels: + params['predict_labels'] = predict_labels + return self._call_create(params, ['requested', 'running']) diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 1af255e6..f2797bb4 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -31,9 +31,9 @@ def undeploy(self, model_id, locally=False): return ModelUndeploy(self.ctx, self.project).execute(model_id, locally) def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, - output=None, no_features_in_result=None, score=False, score_true_data=None): + output=None, no_features_in_result=None, score=False, score_true_data=None, predict_labels=None): return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, - predicted_at, output, no_features_in_result, score, score_true_data) + predicted_at, output, no_features_in_result, score, score_true_data, predict_labels) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, experiment_params=None, locally=False): diff --git a/a2ml/api/auger/impl/mparts/predict.py b/a2ml/api/auger/impl/mparts/predict.py index d523f6ca..8743c065 100644 --- a/a2ml/api/auger/impl/mparts/predict.py +++ b/a2ml/api/auger/impl/mparts/predict.py @@ -24,7 +24,7 @@ def __init__(self, ctx): def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None, - score=False, score_true_data=None): + score=False, score_true_data=None, predict_labels=None): if filename is not None and isinstance(filename, str) and \ not (filename.startswith("http:") or filename.startswith("https:")) and \ not fsclient.is_s3_path(filename): @@ -34,13 +34,13 @@ def execute(self, filename, model_id, threshold=None, locally=False, data=None, if locally: if locally == "docker": predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) else: predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) else: predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) return predicted @@ -104,7 +104,7 @@ def _check_model_project(self, pipeline_api): self.ctx.config.get('name'), model_project_name)) def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, - output, no_features_in_result, score, score_true_data): + output, no_features_in_result, score, score_true_data, predict_labels): records, features, file_url, is_pandas_df = self._process_input(filename, data, columns) temp_file = None ds_result = None @@ -114,7 +114,7 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic pipeline_api = AugerPipelineApi(self.ctx, None, model_id) predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at, no_features_in_result=no_features_in_result, - score=score, score_true_data=score_true_data) + score=score, score_true_data=score_true_data, predict_labels=predict_labels) try: ds_result = DataFrame.create_dataframe(predictions.get('signed_prediction_url'), @@ -145,7 +145,7 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic fsclient.remove_file(temp_file) def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, - output, no_features_in_result, score, score_true_data): + output, no_features_in_result, score, score_true_data, predict_labels): from auger_ml.model_exporter import ModelExporter is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) @@ -161,12 +161,17 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre if score and score_true_data is None: options = fsclient.read_json_file(os.path.join(model_path, "options.json")) - ds = DataFrame.create_dataframe(filename_arg, data, [options['targetFeature']]) + ds = DataFrame.create_dataframe(filename_arg, data)#, [options['targetFeature']]) score_true_data = ds.df - - res, options = ModelExporter({}).predict_by_model_to_ds(model_path, - path_to_predict=filename_arg, records=data, features=columns, - threshold=threshold, no_features_in_result=no_features_in_result) + + if predict_labels: + res, options = ModelExporter({}).predict_labels_by_model_to_ds(model_path, + path_to_predict=filename_arg, records=data, features=columns, + threshold=threshold, no_features_in_result=no_features_in_result, predict_labels=predict_labels) + else: + res, options = ModelExporter({}).predict_by_model_to_ds(model_path, + path_to_predict=filename_arg, records=data, features=columns, + threshold=threshold, no_features_in_result=no_features_in_result) ds_result = DataFrame({'data_path': None}) ds_result.df = res.df @@ -192,7 +197,7 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre # no_features_in_result=no_features_in_result) #, output=output) def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, - output, no_features_in_result, score, score_true_data): + output, no_features_in_result, score, score_true_data, predict_labels): model_deploy = ModelDeploy(self.ctx, None) is_model_loaded, model_path = model_deploy.verify_local_model(model_id, add_model_folder=False) if not is_model_loaded: @@ -205,7 +210,7 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co filename = os.path.join(self.ctx.config.get_path(), '.augerml', 'predict_data.csv') ds.saveToCsvFile(filename, compression=None) - predicted = self._docker_run_predict(filename, threshold, model_path, score, score_true_data) + predicted = self._docker_run_predict(filename, threshold, model_path, score, score_true_data, predict_labels) if not filename_arg: ds_result = DataFrame.create_dataframe(predicted) @@ -223,7 +228,7 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co return predicted - def _docker_run_predict(self, filename, threshold, model_path, score, score_true_data): + def _docker_run_predict(self, filename, threshold, model_path, score, score_true_data, predict_labels): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) docker_tag = cluster_settings.get('kubernetes_stack') predict_file = os.path.basename(filename) diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 326a219f..5f061718 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -23,15 +23,15 @@ def deploy(self, project, model_id, locally, review, name, algorithm, score, dat @authenticated #@with_project(autocreate=False) def predict(self, filename, model_id, threshold, locally, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data): + no_features_in_result, score, score_true_data, predict_labels): if locally: self.deploy(model_id, locally, review=False, name=None, algorithm=None, score=None, data_path=None) predicted = Model(self.ctx, project=None).predict( filename, model_id, threshold, locally, data, columns, predicted_at, output, - no_features_in_result, score, score_true_data) + no_features_in_result, score, score_true_data, predict_labels) - if filename: + if output: self.ctx.log('Predictions stored in %s' % predicted) if isinstance(predicted, dict) and 'predicted' in predicted: diff --git a/a2ml/api/utils/formatter.py b/a2ml/api/utils/formatter.py index e7a118bb..d5f37508 100644 --- a/a2ml/api/utils/formatter.py +++ b/a2ml/api/utils/formatter.py @@ -12,7 +12,8 @@ def print_table(log, table_list, headers=None, hor_lines=True): col_list = list(table_list[0].keys() if table_list else []) row_list = [col_list] # 1st row = header for item in table_list: - row_list.append([str(item.get(col) or '') for col in col_list]) + row_list.append([str(item.get(col)) if item.get(col) is not None else '' for col in col_list]) + # maximun size of the col for each element col_size = [max(map(len, col)) for col in zip(*row_list)] # insert seperating line before every line, and extra one for ending. diff --git a/setup.py b/setup.py index 646d8b7c..3998fdeb 100644 --- a/setup.py +++ b/setup.py @@ -84,13 +84,13 @@ def run(self): 'google-cloud-automl' ], 'predict': [ - 'auger.ai.predict[all]==1.0.104' + 'auger.ai.predict[all]==1.0.106' ], 'predict_no_cat_lgbm': [ - 'auger.ai.predict[no_cat_lgbm]==1.0.104' + 'auger.ai.predict[no_cat_lgbm]==1.0.106' ], 'predict_no_lgbm': [ - 'auger.ai.predict[no_cat_lgbm]==1.0.104', + 'auger.ai.predict[no_cat_lgbm]==1.0.106', 'catboost' ] }

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2022-09-29T10:02:12

0.0

[]

augerai/a2ml

augerai__a2ml-610

0198cb1d42783000318a5e4579018a8be362743f

diff --git a/.circleci/config.yml b/.circleci/config.yml deleted file mode 100644 index 8b70a2d2..00000000 --- a/.circleci/config.yml +++ /dev/null @@ -1,254 +0,0 @@ -version: 2.1 - -orbs: - docker: circleci/[email protected] - aws-eks: circleci/[email protected] - role: airswap/[email protected] - aws-cli: circleci/[email protected] - kubernetes: circleci/[email protected] - helm: circleci/[email protected] - -jobs: - build-docs: - docker: - - image: circleci/python:3.7-stretch - steps: - - checkout - - run: sudo chown -R circleci:circleci /usr/local/bin - - run: sudo chown -R circleci:circleci /usr/local/lib/python3.7/site-packages - - restore_cache: - keys: - - a2ml-python-doc-deps-v3-{{ arch }}-3.7-{{ .Branch }}-{{ checksum "setup.py" }}-{{ checksum "docs/requirements.txt" }} - - a2ml-python-doc-deps-v3-{{ arch }}-3.7-{{ .Branch }} - - a2ml-python-doc-deps-v3-{{ arch }}-3.7 - - run: - name: Install dependencies - command: | - virtualenv venv - source venv/bin/activate - make develop-docs - - save_cache: - key: a2ml-python-doc-deps-v3-{{ arch }}-3.7-{{ .Branch }}-{{ checksum "setup.py" }}-{{ checksum "docs/requirements.txt" }} - paths: - - "venv" - - "/home/circleci/.cache/pip" - - run: - name: Build docs - command: | - source venv/bin/activate - cd docs/ - make html - - persist_to_workspace: - root: docs/build - paths: html - publish-docs: - docker: - - image: node:10.15.0 - steps: - - checkout - - attach_workspace: - at: docs/build - - add_ssh_keys: - fingerprints: "44:aa:23:95:60:12:6b:b5:8d:b2:e5:05:24:1f:94:cf" - - run: - name: Deploy docs to gh-pages branch - command: | - git config user.email "[email protected]" - git config user.name "augerbot" - npm install -g --silent [email protected] - gh-pages --dotfiles --message "[skip ci] Updates" --dist docs/build/html - build-and-test: - docker: - - image: cimg/base:stable - steps: - - checkout - - setup_remote_docker - - run: - name: Install Docker Compose - environment: - COMPOSE_VERSION: '1.29.2' - command: | - curl -L "https://github.com/docker/compose/releases/download/${COMPOSE_VERSION}/docker-compose-$(uname -s)-$(uname -m)" -o ~/docker-compose - chmod +x ~/docker-compose - sudo mv ~/docker-compose /usr/local/bin/docker-compose - - run: make config docker-test - - run: make docker-save - - persist_to_workspace: - root: . - paths: - - ./image.tar.gz - publish-docker: - executor: docker/machine - parameters: - docker-tag: - description: Tag to publish - type: string - default: latest - steps: - - checkout - - attach_workspace: - at: ./ - - docker/check - - run: make docker-load - - run: DOCKER_TAG=<< parameters.docker-tag >> make docker-tag - - docker/push: - image: augerai/a2ml - tag: << parameters.docker-tag >> - - publish-pip: - docker: - - image: circleci/python:3.7-stretch - steps: - - checkout - - run: - command: | - echo -e "[pypi]" >> ~/.pypirc - echo -e "username = $PYPI_USERNAME" >> ~/.pypirc - echo -e "password = $PYPI_PASSWORD" >> ~/.pypirc - - run: make build - - run: make release - - deploy-to-k8s: - executor: aws-eks/python3 - parameters: - cluster-name: - description: | - Name of the EKS cluster - type: string - default: ${STAGING_CLUSTER_NAME} - aws-region: - description: | - AWS region - type: string - default: ${AWS_DEFAULT_REGION} - namespace: - description: | - a2ml namespace - type: string - default: a2ml - release-name: - description: | - a2ml helm release-name - type: string - default: a2ml - reuse-values: - description: | - Reuse last release's values and merge in any overrides - type: boolean - default: true - account-id: - description: | - AWS account containing the cluster - type: string - default: ${ACCOUNT_ID} - role-name: - description: | - AWS role to assume for deploying to eks - type: string - default: ${ROLE_NAME} - values-to-override: - description: | - Values will be used as helm install --set "key1=value1,key2=value2" - type: string - default: "" - steps: - - aws-cli/setup - - role/assume-role: - account-id: << parameters.account-id >> - role-name: << parameters.role-name >> - - aws-eks/update-kubeconfig-with-authenticator: - cluster-name: << parameters.cluster-name >> - aws-region: << parameters.aws-region >> - install-kubectl: true - - helm/install-helm-client: - version: v3.4.2 - - run: - command: | - helm repo add augerai https://augerai.github.io/charts - helm repo update - helm repo list - name: Add augerai repo - - helm/upgrade-helm-chart: - helm-version: v3.2.4 - chart: augerai/a2ml - namespace: << parameters.namespace >> - release-name: << parameters.release-name >> - reuse-values: << parameters.reuse-values >> - values-to-override: << parameters.values-to-override >> - -workflows: - build-test-publish: - jobs: - - build-and-test: - filters: - tags: - only: /^v.*/ - - build-docs - - publish-docs: - requires: - - build-and-test - - build-docs - filters: - branches: - only: - - master - - publish-docker: - context: docker-hub - name: docker-publish-tag - docker-tag: $CIRCLE_TAG - requires: - - build-and-test - filters: - tags: - only: /^v.*/ - branches: - ignore: /.*/ - - publish-docker: - context: docker-hub - name: docker-publish-master - requires: - - build-and-test - filters: - branches: - only: - - master - tags: - ignore: /^v.*/ - - deploy-to-k8s: - name: deploy-to-k8s-master - cluster-name: ${STAGING_CLUSTER_NAME} - values-to-override: "image.tag=latest" - release-name: a2ml - namespace: a2ml - context: eks - requires: - - docker-publish-master - filters: - branches: - only: - - master - tags: - ignore: /^v.*/ - - deploy-to-k8s: - name: deploy-to-k8s - namespace: a2ml - release-name: a2ml - values-to-override: "image.tag=$CIRCLE_TAG" - context: eks - requires: - - docker-publish-tag - account-id: ${PROD_ACCOUNT_ID} - cluster-name: ${PROD_CLUSTER_NAME} - role-name: ${PROD_ROLE_NAME} - filters: - tags: - only: /^v.*/ - branches: - ignore: /.*/ - - publish-pip: - filters: - tags: - only: /^v.*/ - branches: - ignore: /.*/ - context: pypi diff --git a/.github/workflows/docs.yml b/.github/workflows/docs.yml new file mode 100644 index 00000000..f87c87d2 --- /dev/null +++ b/.github/workflows/docs.yml @@ -0,0 +1,32 @@ +name: docs + +on: + push: + tags: + - v* + +jobs: + build_publish_docs: + runs-on: ubuntu-latest + + steps: + - uses: actions/checkout@v3 + - name: Set up Python + uses: actions/setup-python@v3 + with: + python-version: 3.7 + + - name: Build docs + run: | + make develop-docs + cd docs/ + make html + + - name: Deploy + uses: peaceiris/actions-gh-pages@v3 + with: + github_token: ${{ secrets.GITHUB_TOKEN }} + publish_dir: docs/build/html + + + \ No newline at end of file diff --git a/.github/workflows/publish_docker.yml b/.github/workflows/publish_docker.yml new file mode 100644 index 00000000..80739507 --- /dev/null +++ b/.github/workflows/publish_docker.yml @@ -0,0 +1,69 @@ +name: publish_docker + +on: + push: + tags: + - v* + branches: + - master + +jobs: + publish_docker: + runs-on: ubuntu-latest + + permissions: + id-token: write + contents: read + env: + DOCKER_TAG: ${{github.ref_type == 'tag' && github.ref_name || 'latest'}} + REPO_NAME: 'augerai/a2ml' + DOCKER_USER: ${{ secrets.DOCKER_USER }} + DOCKER_PASS: ${{ secrets.DOCKER_PASS }} + + AWS_KS_ROLE: ${{ secrets[format('AWS_KS_{0}_ROLE', github.ref_type == 'tag' && 'STABLE' || 'EXPERIMENTAL')] }} + CLUSTER_NAME: ${{ secrets[format('{0}_CLUSTER_NAME', github.ref_type == 'tag' && 'STABLE' || 'EXPERIMENTAL')] }} + KUBECONFIG_FILE: '/home/runner/.kube/config' + RELEASE_NAME: 'a2ml' + + steps: + - uses: actions/checkout@v3 + + - name: Build docker and run tests + run: make config docker-test + + - name: Push docker image + run: | + docker login -u $DOCKER_USER -p $DOCKER_PASS + docker push $REPO_NAME:$DOCKER_TAG + + - name: configure aws credentials with role1 + uses: aws-actions/configure-aws-credentials@v1 + with: + role-to-assume: ${{ secrets.AWS_ROLE }} + aws-region: us-west-2 + + - name: Assume execution role + uses: aws-actions/configure-aws-credentials@v1 + with: + aws-access-key-id: ${{ env.AWS_ACCESS_KEY_ID }} + aws-region: us-west-2 + aws-secret-access-key: ${{ env.AWS_SECRET_ACCESS_KEY }} + aws-session-token: ${{ env.AWS_SESSION_TOKEN }} + role-duration-seconds: 3000 + role-skip-session-tagging: true + role-to-assume: ${{ env.AWS_KS_ROLE }} + + - uses: azure/[email protected] + - uses: azure/setup-helm@v1 + with: + version: 'v3.4.2' + + - name: setup the kubectl config + run : aws eks update-kubeconfig --name $CLUSTER_NAME + + - name: Helm upgrade augerai repo + run: | + helm repo add augerai https://augerai.github.io/charts + helm repo update + helm repo list + helm upgrade $RELEASE_NAME $REPO_NAME --namespace=$RELEASE_NAME --set=image.tag=$DOCKER_TAG --reuse-values --wait --atomic --kubeconfig $KUBECONFIG_FILE diff --git a/.github/workflows/publish_pip.yml b/.github/workflows/publish_pip.yml new file mode 100644 index 00000000..e10e2f9a --- /dev/null +++ b/.github/workflows/publish_pip.yml @@ -0,0 +1,33 @@ +name: publish_pip + +on: + push: + tags: + - v* + +jobs: + publish_pip: + runs-on: ubuntu-latest + env: + PYPI_USERNAME: ${{ secrets.PYPI_USERNAME }} + PYPI_PASSWORD: ${{ secrets.PYPI_PASSWORD }} + + steps: + - uses: actions/checkout@v3 + - name: Set up Python + uses: actions/setup-python@v3 + with: + python-version: 3.7 + + - name: Make pypirc file + run: | + echo -e "[pypi]" >> ~/.pypirc + echo -e "username = $PYPI_USERNAME" >> ~/.pypirc + echo -e "password = $PYPI_PASSWORD" >> ~/.pypirc + + - name: Install dependencies + run: pip install wheel + - name: Build wheel + run: make build + - name: Deploy package + run: make release diff --git a/a2ml/__init__.py b/a2ml/__init__.py index 53a7c762..461d9678 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.76' +__version__ = '1.0.77' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index dc72609a..03454e01 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -8,7 +8,7 @@ class A2ML(BaseA2ML): """Facade to A2ML providers.""" def __init__(self, ctx, provider = None): - """Initializes A2ML PREDIT instance. + """Initializes new A2ML PREDIT instance. Args: ctx (object): An instance of the a2ml Context. diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index f84dfa20..11f95335 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -79,6 +79,8 @@ def _do_score_actual(self, df_data, predicted_feature=None, extra_features=[]): def validate_roi_syntax(self, expressions, features=[]): res = [] + logging.info('validate_roi_syntax with experession: %s'%(expressions)) + known_vars = ["A", "P", "$" + self.target_feature] + list( map(lambda name: "$" + name, set(self.original_features + features)) ) diff --git a/docs/requirements.txt b/docs/requirements.txt index 51d880e1..0f8a6e5f 100644 --- a/docs/requirements.txt +++ b/docs/requirements.txt @@ -1,2 +1,2 @@ -sphinx~=3.0.4 +sphinx~=4.4.0 git+https://github.com/augerai/sphinx_rtd_theme.git@bump-version#2ab38df0d303163e0e6c2bac80d907e9915000cb'

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2022-05-09T20:42:48

0.0

[]

augerai/a2ml

augerai__a2ml-607

2273c754ddc486783b6b1095e15c5b67e7c48242

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index a53afbe9..218e3516 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.65' +__version__ = '1.0.66' diff --git a/a2ml/api/auger/config.py b/a2ml/api/auger/config.py index 0a9e47a5..3d8013fb 100644 --- a/a2ml/api/auger/config.py +++ b/a2ml/api/auger/config.py @@ -3,7 +3,7 @@ def __init__(self, ctx): super(AugerConfig, self).__init__() self.ctx = ctx - def set_data_set(self, name, source=None, validation=False): + def set_data_set(self, name, source=None, validation=False, user_name=None): #TODO: add more providers later if validation: self.ctx.config.set('experiment/validation_dataset', name) @@ -12,6 +12,9 @@ def set_data_set(self, name, source=None, validation=False): else: #print("set_data_set: %s"%self.ctx.use_auger_cloud()) self.ctx.config.set('dataset', name) + if user_name: + self.ctx.config.set('dataset_name', user_name) + if self.ctx.use_auger_cloud() and 'azure' in self.ctx.get_providers(): self.ctx.config.set('dataset', name, "azure") @@ -21,9 +24,34 @@ def set_data_set(self, name, source=None, validation=False): def set_experiment(self, experiment_name, experiment_session_id): self.ctx.config.set('experiment/name', experiment_name) self.ctx.config.set('experiment/experiment_session_id', experiment_session_id) + + if self.ctx.config.get('dataset_name'): + dataset_name = self.ctx.config.get('dataset_name') + self.ctx.config.set(f'experiments/{dataset_name}/experiment_id', experiment_name) + self.ctx.config.set(f'experiments/{dataset_name}/experiment_session_id', experiment_session_id) + self.ctx.config.write() + def _get_experiment_by_dataset(self): + dataset_name = self.ctx.config.get('dataset_name') + experiments = self.ctx.config.get('experiments', {}) + + return experiments.get(dataset_name, {}) + + def get_experiment(self): + return self._get_experiment_by_dataset().get('experiment_id', + self.ctx.config.get('experiment/name')) + + def get_experiment_session(self): + return self._get_experiment_by_dataset().get('experiment_session_id', + self.ctx.config.get('experiment/experiment_session_id')) + + def get_dataset(self): + return self._get_experiment_by_dataset().get('dataset_id', + self.ctx.config.get('dataset')) + def set_project(self, project_name): self.ctx.config.set('name', project_name) self.ctx.config.write() return self + diff --git a/a2ml/api/auger/dataset.py b/a2ml/api/auger/dataset.py index 65ec91ea..a99cc098 100644 --- a/a2ml/api/auger/dataset.py +++ b/a2ml/api/auger/dataset.py @@ -18,7 +18,7 @@ def __init__(self, ctx): @with_project(autocreate=False) def list(self, project): count = 0 - selected = self.ctx.config.get('dataset', None) + selected = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) for dataset in iter(DataSet(self.ctx, project).list()): self.ctx.log( ('[%s] ' % ('x' if selected == dataset.get('name') else ' ')) + @@ -40,7 +40,7 @@ def _create(self, project, source = None, validation=False, name=None, descripti if source is None: source = self.ctx.config.get('source', None) dataset = DataSet(self.ctx, project).create(source, name, description) - AugerConfig(self.ctx).set_data_set(dataset.name, source, validation) + AugerConfig(self.ctx).set_data_set(dataset.name, source, validation, name) return dataset @@ -60,7 +60,7 @@ def upload(self, project, source = None, name=None): @with_project(autocreate=False) def delete(self, project, name): if name is None: - name = self.ctx.config.get('dataset', None) + name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) DataSet(self.ctx, project, name).delete() if name == self.ctx.config.get('dataset', None): AugerConfig(self.ctx).set_data_set(None, None, False).set_experiment(None, None) @@ -69,7 +69,7 @@ def delete(self, project, name): @error_handler def select(self, name): - old_name = self.ctx.config.get('dataset', None) + old_name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) if name != old_name: AugerConfig(self.ctx).set_data_set(name, None, False).set_experiment(None, None) self.ctx.log('Selected DataSet %s' % name) @@ -80,7 +80,7 @@ def select(self, name): @with_project(autocreate=False) def download(self, project, name, path_to_download): if name is None: - name = self.ctx.config.get('dataset', None) + name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) file_name = DataSet(self.ctx, project, name).download(path_to_download) self.ctx.log('Downloaded dataset %s to %s' % (name, file_name)) return {'dowloaded': name, 'file': file_name} diff --git a/a2ml/api/auger/experiment.py b/a2ml/api/auger/experiment.py index edb4df07..04dbc3cc 100644 --- a/a2ml/api/auger/experiment.py +++ b/a2ml/api/auger/experiment.py @@ -31,7 +31,7 @@ def list(self, dataset): @with_dataset def start(self, dataset): experiment_name = \ - self.ctx.config.get('experiment/name', None) + AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) experiment_name, session_id = \ Experiment(self.ctx, dataset, experiment_name).start() AugerConfig(self.ctx).set_experiment(experiment_name, session_id) @@ -41,12 +41,12 @@ def start(self, dataset): @authenticated @with_dataset def stop(self, dataset, run_id = None): - name = self.ctx.config.get('experiment/name', None) + name = AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) if name is None: raise AugerException('Please specify Experiment name...') if run_id is None: - run_id = self.ctx.config.get( - 'experiment/experiment_session_id', None) + run_id = AugerConfig(self.ctx).get_experiment_session() + #self.ctx.config.get('experiment/experiment_session_id', None) if Experiment(self.ctx, dataset, name).stop(run_id): self.ctx.log('Search is stopped...') @@ -58,12 +58,13 @@ def stop(self, dataset, run_id = None): @authenticated @with_dataset def leaderboard(self, dataset, run_id = None): - name = self.ctx.config.get('experiment/name', None) + name = AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) if name is None: raise AugerException('Please specify Experiment name...') if run_id is None: - run_id = self.ctx.config.get( - 'experiment/experiment_session_id', None) + run_id = AugerConfig(self.ctx).get_experiment_session() + # run_id = self.ctx.config.get( + # 'experiment/experiment_session_id', None) leaderboard, status, run_id, trials_count, errors = Experiment( self.ctx, dataset, name).leaderboard(run_id) if leaderboard is None: @@ -109,7 +110,7 @@ def leaderboard(self, dataset, run_id = None): @authenticated @with_dataset def history(self, dataset): - name = self.ctx.config.get('experiment/name', None) + name = AugerConfig(self.ctx).get_experiment() #self.ctx.config.get('experiment/name', None) if name is None: raise AugerException('Please specify Experiment name...') for exp_run in iter(Experiment(self.ctx, dataset, name).history()): diff --git a/a2ml/api/auger/impl/decorators.py b/a2ml/api/auger/impl/decorators.py index aaa73523..f681cad2 100644 --- a/a2ml/api/auger/impl/decorators.py +++ b/a2ml/api/auger/impl/decorators.py @@ -33,10 +33,11 @@ def wrapper(self, *args, **kwargs): def with_dataset(decorated): from .dataset import DataSet + from ..config import AugerConfig def wrapper(self, *args, **kwargs): project = _get_project(self, False) - data_set_name = self.ctx.config.get('dataset', None) + data_set_name = AugerConfig(self.ctx).get_dataset() #self.ctx.config.get('dataset', None) if data_set_name is None: raise AugerException( 'Please specify dataset name in auger.yaml/dataset...') diff --git a/a2ml/api/utils/context.py b/a2ml/api/utils/context.py index 853aa570..cab0cf6c 100644 --- a/a2ml/api/utils/context.py +++ b/a2ml/api/utils/context.py @@ -14,7 +14,7 @@ class Context(object): """The Context class provides an environment to run A2ML""" - def __init__(self, name='config', path=None, debug=False): + def __init__(self, name='auger', path=None, debug=False): """Initializes the Context instance Args: @@ -39,7 +39,7 @@ def __init__(self, name='config', path=None, debug=False): self.provider_info = None if len(self.name) > 0: - self.name = "{:<9}".format('[%s]' % self.name) + self.name = f'[{self.name}] ' #"{:<9}".format('[%s]' % self.name) self.debug = self.config.get('debug', debug) self.set_runs_on_server(False) @@ -109,6 +109,8 @@ def is_external_provider(self): return providers and providers[0] == 'external' def copy(self, name): + return self + """creates a copy of an existing Context Args: @@ -123,23 +125,23 @@ def copy(self, name): ctx = Context() new_ctx = ctx.copy() """ - new = Context(name, self.config.path, self.debug) - new.set_runs_on_server(self._runs_on_server) - new.notificator = self.notificator - new.request_id = self.request_id - new.config.parts = self.config.parts - new.config.parts_changes = self.config.parts_changes - - try: - new.config.set("providers", name, config_name='config') - except Exception as e: - # In case if command run in folder without config, do not set it - pass + # new = Context(name, self.config.path, self.debug) + # new.set_runs_on_server(self._runs_on_server) + # new.notificator = self.notificator + # new.request_id = self.request_id + # new.config.parts = self.config.parts + # new.config.parts_changes = self.config.parts_changes + + # try: + # new.config.set("providers", name, config_name='config') + # except Exception as e: + # # In case if command run in folder without config, do not set it + # pass - if hasattr(self, 'credentials'): - new.credentials = self.credentials + # if hasattr(self, 'credentials'): + # new.credentials = self.credentials - return new + # return new def log(self, msg, *args, **kwargs): log.info('%s%s' %(self.name, msg), *args, **kwargs) diff --git a/a2ml/cmdl/commands/cmd_import.py b/a2ml/cmdl/commands/cmd_import.py index 2502ef75..fcc87dbc 100644 --- a/a2ml/cmdl/commands/cmd_import.py +++ b/a2ml/cmdl/commands/cmd_import.py @@ -5,12 +5,14 @@ @click.command('import', short_help='Import data for training.') @click.option('--source', '-s', type=click.STRING, required=False, help='Source file to import.If skipped, then import source from config.yml.') [email protected]('--name', '-n', type=click.STRING, required=False, + help='Name file to import.') @click.option('--description', '-d', type=click.STRING, required=False, help='Description of dataset.') @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') @pass_context -def cmdl(ctx, source, description, provider): +def cmdl(ctx, source, name, description, provider): """Import data for training.""" ctx.setup_logger(format='') - A2ML(ctx, provider).import_data(source, description=description) + A2ML(ctx, provider).import_data(source, name, description=description)

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2022-03-17T09:51:24

0.0

[]

augerai/a2ml

augerai__a2ml-601

bb141e98e7cb89a5d6aa226cb98adbb1b3163add

diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 69f5b188..8424c81a 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -365,7 +365,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at score, score_true_data ) @show_result - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False, provider=None): """Submits actual results(ground truths) for predictions of a deployed model. This is used to review and monitor active models. Note: @@ -396,6 +397,12 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N columns(list): list of column names if data is array of records actuals_at: Actuals date. Use for review of historical data. actual_date_column(str): name of column in data which contains actual date + experiment_params(dict): parameters to calculate experiment metrics :: + + start_date(date): experiment actuals start date + end_date(date): experiment actuals end date + date_col(str): column name with date + locally(bool): Process actuals locally. provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. @@ -437,7 +444,8 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N A2ML(ctx, "external").actuals('external_model_id', data=actual_records,columns=columns) """ - return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', model_id, filename, data, columns, actuals_at, actual_date_column, locally) + return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', + model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) @show_result diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index aaf1cf0c..86dae92b 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -155,7 +155,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at score, score_true_data ) @show_result - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False, provider=None): """Submits actual results(ground truths) for predictions of a deployed model. This is used to review and monitor active models. Note: @@ -186,6 +187,13 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N columns(list): list of column names if data is array of records actuals_at: Actuals date. Use for review of historical data. actual_date_column(str): name of column in data which contains actual date + experiment_params(dict): parameters to calculate experiment metrics :: + + start_date(date): experiment actuals start date + end_date(date): experiment actuals end date + date_col(str): column name with date + + locally(bool): Process actuals locally. provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. @@ -227,7 +235,8 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N A2MLModel(ctx, "external").actuals('external_model_id', data=actual_records,columns=columns) """ - return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', model_id, filename, data, columns, actuals_at, actual_date_column, locally) + return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', + model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) @show_result def review_alert(self, model_id, parameters = None, locally=False, provider=None, name=None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 27c75909..46f7db9d 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -30,9 +30,10 @@ def predict(self, model_id, filename, threshold=None, locally=False, data=None, model_id, filename, threshold, locally, data, columns, predicted_at, output, no_features_in_result, score, score_true_data) - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False): return AugerModel(self.ctx).actuals( - model_id, filename, data, columns, actuals_at, actual_date_column, locally) + model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) def delete_actuals(self, model_id, with_predictions=False, begin_date=None, end_date=None, locally=False): return AugerModel(self.ctx).delete_actuals( diff --git a/a2ml/api/auger/impl/cloud/actual.py b/a2ml/api/auger/impl/cloud/actual.py index 17e08373..c29c7f23 100644 --- a/a2ml/api/auger/impl/cloud/actual.py +++ b/a2ml/api/auger/impl/cloud/actual.py @@ -12,7 +12,7 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self.parent_id_name = "endpoint_id" self._set_api_request_path("AugerEndpointActualApi") - def create(self, records, features, actuals_at, actuals_path, actual_date_column): + def create(self, records, features, actuals_at, actuals_path, actual_date_column, experiment_params): params = {} if self.use_endpoint: params['endpoint_id'] = self.parent_api.object_id @@ -35,6 +35,10 @@ def create(self, records, features, actuals_at, actuals_path, actual_date_column if actuals_at: params['actuals_at'] = str(actuals_at) + if experiment_params: + params['experiment_params'] = experiment_params + + print(params) return self._call_create( params=params, has_return_object=False) diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index 4c2e4605..b5c1847e 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -94,12 +94,12 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at return prediction_properties.get('result') - def actual(self, records, features, actuals_at, actuals_path, actual_date_column): + def actual(self, records, features, actuals_at, actuals_path, actual_date_column, experiment_params): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') actual_api = AugerActualApi(self.ctx, self, use_endpoint=self.check_endpoint()) - actual_api.create(records, features, actuals_at, actuals_path, actual_date_column) + actual_api.create(records, features, actuals_at, actuals_path, actual_date_column, experiment_params) #TODO: get object actual from cloud return True diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 89e2b25e..72826d08 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -35,7 +35,8 @@ def predict(self, filename, model_id, threshold=None, locally=False, data=None, return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result, score, score_true_data) - def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): + def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False): if locally: is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) @@ -57,10 +58,12 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N columns=columns, actual_date=actuals_at, actual_date_column=actual_date_column, + experiment_params=experiment_params, provider='auger' ) else: - return ModelActual(self.ctx).execute(model_id, filename, data, columns, actuals_at, actual_date_column) + return ModelActual(self.ctx).execute(model_id, filename, data, columns, actuals_at, + actual_date_column, experiment_params) def delete_actuals(self, model_id, with_predictions=False, begin_date=None, end_date=None, locally=False): if locally: diff --git a/a2ml/api/auger/impl/mparts/actual.py b/a2ml/api/auger/impl/mparts/actual.py index aad79269..50edcb7b 100644 --- a/a2ml/api/auger/impl/mparts/actual.py +++ b/a2ml/api/auger/impl/mparts/actual.py @@ -12,9 +12,9 @@ def __init__(self, ctx): super(ModelActual, self).__init__() self.ctx = ctx - def execute(self, model_id, filename, data, columns, actuals_at, actual_date_column): + def execute(self, model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params): records, features, file_url, is_pandas_df = ModelPredict(self.ctx)._process_input(filename, data, columns=columns) pipeline_api = AugerPipelineApi(self.ctx, None, model_id) - return pipeline_api.actual(records, features, actuals_at, file_url, actual_date_column) + return pipeline_api.actual(records, features, actuals_at, file_url, actual_date_column, experiment_params) diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 7901f871..326a219f 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -42,8 +42,9 @@ def predict(self, filename, model_id, threshold, locally, data, columns, predict @error_handler @authenticated @with_project(autocreate=False) - def actuals(self, project, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): - return Model(self.ctx, project).actuals(model_id, filename, data, columns, actuals_at, actual_date_column, locally) + def actuals(self, project, model_id, filename=None, data=None, columns=None, actuals_at=None, + actual_date_column=None, experiment_params=None, locally=False): + return Model(self.ctx, project).actuals(model_id, filename, data, columns, actuals_at, actual_date_column, experiment_params, locally) @error_handler @authenticated diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index 6df5be05..1d2e0bf9 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -145,7 +145,7 @@ def add_external_model(self, target_column, scoring, task_type, binary_classific def add_actuals( self, ctx, actuals_path=None, data=None, columns=None, external_model=False, actual_date=None, actual_date_column=None, actuals_id = None, return_count=False, provider='auger', - do_predict=False + do_predict=False, experiment_params=None ): ds_actuals = DataFrame.create_dataframe(actuals_path, data, features=columns) @@ -178,8 +178,12 @@ def add_actuals( result = self._do_score_actual(ds_actuals.df) baseline_score = {} + experiment_score = {} + experiment_count = 0 if "baseline_target" in ds_actuals.columns: baseline_score = self._do_score_actual(ds_actuals.df, "baseline_target") + if experiment_params: + experiment_score, experiment_count = self._do_score_actual_experiment(ds_actuals, experiment_params) #logging.info("Actual result: %s", result) ds_actuals.df = ds_actuals.df.rename(columns={self.target_feature: 'a2ml_predicted'}) @@ -207,10 +211,34 @@ def add_actuals( ds_actuals.saveToFeatherFile(os.path.join(self.model_path, "predictions", file_name)) if return_count: - return {'score': result, 'count': actuals_count, 'baseline_score': baseline_score} + return {'score': result, 'count': actuals_count, 'baseline_score': baseline_score, + 'experiment_score': experiment_score, 'experiment_count': experiment_count} else: return result + def _do_score_actual_experiment(self, ds_actuals, experiment_params): + if experiment_params.get('start_date') and experiment_params.get('end_date'): + df_exp_actuals = ds_actuals.df.query("%s>='%s' and %s<'%s'"%( + experiment_params.get('date_col'), + experiment_params.get('start_date'), + experiment_params.get('date_col'), + experiment_params.get('end_date') + )) + elif experiment_params.get('start_date'): + df_exp_actuals = ds_actuals.df.query("%s>='%s'"%( + experiment_params.get('date_col'), + experiment_params.get('start_date') + )) + elif experiment_params.get('end_date'): + df_exp_actuals = ds_actuals.df.query("%s<'%s'"%( + experiment_params.get('date_col'), + experiment_params.get('end_date') + )) + else: + df_exp_actuals = ds_actuals.df + + return self._do_score_actual(df_exp_actuals), len(df_exp_actuals) + def _do_predict(self, ctx, ds_actuals, provider, predict_feature=None, predicted_at=None): missing_features = set(self.original_features) - set(ds_actuals.columns) if len(missing_features) > 0: diff --git a/a2ml/tasks_queue/tasks_hub_api.py b/a2ml/tasks_queue/tasks_hub_api.py index 73bafa16..d5ccbae4 100644 --- a/a2ml/tasks_queue/tasks_hub_api.py +++ b/a2ml/tasks_queue/tasks_hub_api.py @@ -560,7 +560,8 @@ def score_actuals_by_model_task(params): return_count=params.get('return_count', False), provider=params.get('provider'), external_model=external_model, - do_predict=params.get('do_predict', False) + do_predict=params.get('do_predict', False), + experiment_params=params.get('experiment_params'), ) @celeryApp.task(ignore_result=True)

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-10-29T17:51:10

0.0

[]

augerai/a2ml

augerai__a2ml-594

c255bd05603d41d4bfb604ec940225d2656b0b6c

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index b41bc4d5..f47e40c5 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.37' +__version__ = '1.0.38' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 14fd78f0..6049297e 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -1,6 +1,6 @@ from a2ml.api.base_a2ml import BaseA2ML from a2ml.api.utils.show_result import show_result -from a2ml.api.utils import convert_source +#from a2ml.api.utils import convert_source from a2ml.api.utils.context import Context @@ -28,7 +28,7 @@ def __init__(self, ctx, provider = None): self.local_runner = lambda: self.build_runner(ctx, provider, force_local=True) @show_result - def import_data(self, source=None): + def import_data(self, source=None, name=None): """Imports data defined in context. Uploading the same file name will result in versions being appended to the file name. Note: @@ -46,6 +46,7 @@ def import_data(self, source=None): Args: source (str, optional): Local file name, remote url to the data source file, Pandas DataFrame or postgres url + name (str, optional): Name of dataset, if none then file name used. If source is DataFrame then name should be specified. Returns: Results for each provider. :: @@ -70,8 +71,7 @@ def import_data(self, source=None): a2ml = A2ML(ctx, 'auger, azure') a2ml.import_data() """ - with convert_source(source, self.ctx.config.get("name", "source_data")) as data_source: - return self.runner.execute('import_data', source=data_source) + return self.runner.execute('import_data', source=source, name=name) @show_result def preprocess_data(self, data, preprocessors, locally=False): diff --git a/a2ml/api/a2ml_dataset.py b/a2ml/api/a2ml_dataset.py index 1a623e87..9861efa2 100644 --- a/a2ml/api/a2ml_dataset.py +++ b/a2ml/api/a2ml_dataset.py @@ -1,6 +1,6 @@ from a2ml.api.base_a2ml import BaseA2ML from a2ml.api.utils.show_result import show_result -from a2ml.api.utils import convert_source +#from a2ml.api.utils import convert_source class A2MLDataset(BaseA2ML): @@ -58,11 +58,12 @@ def list(self): return self.runner.execute('list') @show_result - def create(self, source = None): + def create(self, source = None, name=None): """Create a new DataSet for the Project specified in the .yaml. Args: source (str, optional): Local file name, remote url to the data source file, Pandas DataFrame or postgres url + name (str, optional): Name of dataset, if none then file name used. If source is DataFrame then name should be specified. Returns: Results for each provider. :: @@ -82,8 +83,7 @@ def create(self, source = None): ctx = Context() dataset = DataSet(ctx, 'auger, azure').create('../dataset.csv') """ - with convert_source(source, self.ctx.config.get("name", "source_data")) as data_source: - return self.runner.execute('create', data_source) + return self.runner.execute('create', data_source, name) @show_result def delete(self, name = None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 394c873f..86cc3b80 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -9,8 +9,8 @@ def __init__(self, ctx): super(AugerA2ML, self).__init__() self.ctx = ctx - def import_data(self, source=None): - return AugerDataset(self.ctx).create(source=source) + def import_data(self, source=None, name=None): + return AugerDataset(self.ctx).create(source=source, name=name) def preprocess_data(self, data, preprocessors, locally=False): return AugerDataset(self.ctx).preprocess_data(data, preprocessors, locally) diff --git a/a2ml/api/auger/dataset.py b/a2ml/api/auger/dataset.py index 4e7ffc6a..0619f62c 100644 --- a/a2ml/api/auger/dataset.py +++ b/a2ml/api/auger/dataset.py @@ -31,15 +31,15 @@ def list(self, project): @error_handler @authenticated @with_project(autocreate=True) - def create(self, project, source = None, validation=False): - dataset = self._create(project, source, validation) + def create(self, project, source = None, validation=False, name=None): + dataset = self._create(project, source, validation, name) self.ctx.log('Created DataSet %s' % dataset.name) return {'created': dataset.name} - def _create(self, project, source = None, validation=False): + def _create(self, project, source = None, validation=False, name=None): if source is None: source = self.ctx.config.get('source', None) - dataset = DataSet(self.ctx, project).create(source) + dataset = DataSet(self.ctx, project).create(source, name) AugerConfig(self.ctx).set_data_set(dataset.name, source, validation) return dataset diff --git a/a2ml/api/auger/impl/cloud/dataset.py b/a2ml/api/auger/impl/cloud/dataset.py index 6419bfda..ae56a0c7 100644 --- a/a2ml/api/auger/impl/cloud/dataset.py +++ b/a2ml/api/auger/impl/cloud/dataset.py @@ -6,6 +6,7 @@ import urllib.parse import urllib.request import xml.etree.ElementTree as ElementTree +from a2ml.api.utils.dataframe import DataFrame from .cluster import AugerClusterApi from .project_file import AugerProjectFileApi @@ -31,8 +32,17 @@ def do_upload_file(self, data_source_file, data_set_name=None, local_data_source # AugerDataSetApi.verify(data_source_file, self.ctx.config.path) if local_data_source: - file_url = self._upload_to_cloud(data_source_file) - file_name = os.path.basename(data_source_file) + if DataFrame.is_dataframe(data_source_file): + with fsclient.save_atomic("%s.parquet"%data_set_name, move_file=False) as local_path: + ds = DataFrame.create_dataframe(data_source_file) + ds.saveToParquetFile(local_path) + file_url = self._upload_to_cloud(local_path) + + file_name = data_set_name + else: + file_url = self._upload_to_cloud(data_source_file) + file_name = os.path.basename(data_source_file) + if data_set_name: self.object_name = data_set_name else: @@ -81,6 +91,9 @@ def _get_readable_name(self): @staticmethod def verify(data_source_file, config_path=None): + if DataFrame.is_dataframe(data_source_file): + return data_source_file, True + if urllib.parse.urlparse(data_source_file).scheme in ['http', 'https']: return data_source_file, False diff --git a/a2ml/api/auger/impl/dataset.py b/a2ml/api/auger/impl/dataset.py index 6a86af1a..b05fb435 100644 --- a/a2ml/api/auger/impl/dataset.py +++ b/a2ml/api/auger/impl/dataset.py @@ -10,7 +10,7 @@ def __init__(self, ctx, project, data_set_name=None): ctx, project, data_set_name) self.project = project - def create(self, data_source_file): + def create(self, data_source_file, name): if data_source_file is None: raise AugerException('Please specify data source file...') @@ -19,7 +19,7 @@ def create(self, data_source_file): self.project.start() - super().create(data_source_file, self.object_name, local_data_source=local_data_source) + super().create(data_source_file, name if name else self.object_name, local_data_source=local_data_source) return self def upload_file(self, data_source_file): diff --git a/a2ml/api/utils/__init__.py b/a2ml/api/utils/__init__.py index 22429b26..6acf1a52 100644 --- a/a2ml/api/utils/__init__.py +++ b/a2ml/api/utils/__init__.py @@ -215,14 +215,14 @@ def convert_to_date(date): else: return date [email protected] -def convert_source(source, name): - if source is not None and isinstance(source, pd.DataFrame): - with fsclient.save_atomic("%s.parquet"%name, move_file=False) as local_path: - source.to_parquet(local_path, index=False, compression="gzip") - yield local_path - else: - yield source +# @contextlib.contextmanager +# def convert_source(source, name): +# if source is not None and isinstance(source, pd.DataFrame): +# with fsclient.save_atomic("%s.parquet"%name, move_file=False) as local_path: +# source.to_parquet(local_path, index=False, compression="gzip") +# yield local_path +# else: +# yield source def retry_helper(func, retry_errors=[], num_try=10, delay=10, ctx=None): nTry = 0 diff --git a/a2ml/api/utils/dataframe.py b/a2ml/api/utils/dataframe.py index ca737052..d81998f7 100644 --- a/a2ml/api/utils/dataframe.py +++ b/a2ml/api/utils/dataframe.py @@ -43,21 +43,48 @@ def _get_compression(self, extension): return compression @staticmethod - def create_dataframe(data_path=None, records=None, features=None): - if data_path: - ds = DataFrame({'data_path': data_path}) - ds.load(features = features) - elif records is not None and isinstance(records, pd.DataFrame): - ds = DataFrame({}) - ds.df = records - if features: - ds.df = ds.df[features] - - ds.from_pandas = True + def create_dataframe(data_path=None, records=None, features=None, reset_index=False): + if data_path is not None: + if isinstance(data_path, pd.DataFrame): + ds = DataFrame({}) + ds.df = data_path + elif isinstance(data_path, DataFrame): + ds = data_path + elif isinstance(data_path, list): + ds = DataFrame({}) + ds.load_records(data_path) + elif isinstance(data_path, dict): + ds = DataFrame({}) + + if 'data' in data_path and 'columns' in data_path: + ds.load_records(data_path['data'], features=data_path['columns']) + else: + ds.load_records(data_path) + else: + ds = DataFrame({'data_path': data_path}) + ds.load(features = features) else: ds = DataFrame({}) ds.load_records(records, features=features) + if reset_index and ds.df is not None: + ds.df.reset_index(drop=True, inplace=True) + + + # if data_path: + # ds = DataFrame({'data_path': data_path}) + # ds.load(features = features) + # elif records is not None and isinstance(records, pd.DataFrame): + # ds = DataFrame({}) + # ds.df = records + # if features: + # ds.df = ds.df[features] + + # ds.from_pandas = True + # else: + # ds = DataFrame({}) + # ds.load_records(records, features=features) + return ds @staticmethod @@ -72,6 +99,10 @@ def load_from_files(files, features=None): except Exception as exc: logging.exception("load_from_files failed for: %s. Error: %s"%(path, exc)) + @staticmethod + def is_dataframe(data): + return isinstance(data, pd.DataFrame) or isinstance(data, DataFrame) + def load_from_file(self, path, features=None, nrows=None): from collections import OrderedDict

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-09-01T21:01:12

0.0

[]

augerai/a2ml

augerai__a2ml-591

e8fc37867e2ede621fa14a6877f661d62de5ae20

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index 1c327d42..24d5b4c5 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.34' +__version__ = '1.0.35' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index aa0e4284..14fd78f0 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -73,6 +73,61 @@ def import_data(self, source=None): with convert_source(source, self.ctx.config.get("name", "source_data")) as data_source: return self.runner.execute('import_data', source=data_source) + @show_result + def preprocess_data(self, data, preprocessors, locally=False): + """Preprocess data + + Args: + data (str|pandas.DataFrame): Input data for preprocess. Can be path to file(local or s3) or Pandas Dataframe + preprocessors (array of dicts): List of preprocessors with parameters :: + + [ + {'text': {'text_cols': []}} + ] + + Preprocessors: + text + * text_cols(array): List of text columns to process + * text_metrics ['mean_length', 'unique_count', 'separation_score'] : Calculate metrics for text fields and after vectorize(separation_score) + * tokenize (dict): Default - {'max_text_len': 30000, 'tokenizers': ['sent'], 'remove_chars': '○•'} + * vectorize ('en_use_lg'|'hashing'|'en_use_md'|'en_use_cmlm_md'|'en_use_cmlm_lg'): See see https://github.com/MartinoMensio/spacy-universal-sentence-encoder + * dim_reduction(dict): Generate features based on vectors. See https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html :: + + { + 'alg_name': 'PCA'|'t-SNE', + 'args': {'n_components': 2} #Number of components to keep. + } + + * output_prefix (str): Prefix for generated columns. Format name: {prefix}_{colname}_{num} + + * calc_distance ['none', 'cosine', 'cityblock', 'euclidean', 'haversine', 'l1', 'l2', 'manhattan', 'nan_euclidean'] | 'cosine' : See https://scikit-learn.org/stable/modules/generated/sklearn.metrics.pairwise.distance_metrics.html#sklearn.metrics.pairwise.distance_metrics + * compare_pairs (array of dicts): When calc_distance is not none. :: + + [ + {'compare_cols': [{'dataset_idx': 0, 'cols': ['col1']}, {'dataset_idx': 1, 'cols': ['col2']}], + 'output_name':'cosine_col1_col2', 'params': {} + }, + {'compare_cols': [{'dataset_idx': 0, 'cols': ['col3']}, {'dataset_idx': 1, 'cols': ['col4']}], + 'output_name':'cosine_col3_col4', 'params': {} + }, + ] + + * datasets: List of datasets to process, may be empty, so all fields takes from main dataset :: + + [ + {'path': 'path', 'keys': ['main_key', 'local_key'], 'text_metrics': ['separation_score', 'mean_length', 'unique_count']}, + {'path': 'path1', 'keys': ['main_key1', 'local_key1']} + ] + + Returns: + { + 'result': True, + 'data': 'data in input format' + } + + """ + return self.get_runner(locally).execute_one_provider('preprocess_data', data, preprocessors, locally) + @show_result def train(self): """Starts training session based on context state. diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 2a55df5e..394c873f 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -12,6 +12,9 @@ def __init__(self, ctx): def import_data(self, source=None): return AugerDataset(self.ctx).create(source=source) + def preprocess_data(self, data, preprocessors, locally=False): + return AugerDataset(self.ctx).preprocess_data(data, preprocessors, locally) + def train(self): return AugerExperiment(self.ctx).start() diff --git a/a2ml/api/auger/dataset.py b/a2ml/api/auger/dataset.py index dc0b6fee..4e7ffc6a 100644 --- a/a2ml/api/auger/dataset.py +++ b/a2ml/api/auger/dataset.py @@ -73,3 +73,25 @@ def download(self, project, name, path_to_download): file_name = DataSet(self.ctx, project, name).download(path_to_download) self.ctx.log('Downloaded dataset %s to %s' % (name, file_name)) return {'dowloaded': name, 'file': file_name} + + def preprocess_data(self, data, preprocessors, locally): + if locally: + return self._preprocess_data_locally(data, preprocessors) + else: + raise Exception("preprocess_data supported with locally=True only.") + + def _preprocess_data_locally(self, data, preprocessors): + from auger_ml.preprocessors.text import TextPreprocessor + + res = data + for p in preprocessors: + name = list(p.keys())[0] + params = list(p.values())[0] + if name != 'text': + raise Exception("Only text preprocessor supported.") + + tp = TextPreprocessor(params) + res = tp.fit_transform(res) + + return res + \ No newline at end of file diff --git a/setup.py b/setup.py index 0a6f572f..2b11a88c 100644 --- a/setup.py +++ b/setup.py @@ -86,7 +86,7 @@ def run(self): 'google-cloud-automl' ], 'predict': [ - 'auger.ai.predict==1.0.79' + 'auger.ai.predict==1.0.80' ] }

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-08-12T20:59:36

0.0

[]

augerai/a2ml

augerai__a2ml-587

858d43f5475354996519821e2f99d8a038440a23

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index 93dae163..0630b51e 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.32' +__version__ = '1.0.33' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 09fe4116..aa0e4284 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -221,7 +221,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, no_features_in_result = None, locally=False, provider=None): + threshold=None, score=False, score_true_data=None, + output=None, no_features_in_result = None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -235,6 +236,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at columns(list): list of column names if data is array of records predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. + score(bool): Calculate scores for predicted results. + score_true_data(str, pandas.DataFrame, dict): Data with true values to calculate scores. If missed, target from filename used for true values. output(str): Output csv file path. no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model @@ -301,7 +304,9 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are stored in rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, + threshold, locally, data, columns, predicted_at, output, no_features_in_result, + score, score_true_data ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 2be596f0..b58e2132 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -66,7 +66,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, no_features_in_result=None, locally=False, provider=None): + threshold=None, score=False, score_true_data=None, + output=None, no_features_in_result=None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -80,6 +81,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at columns(list): list of column names if data is array of records predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. + score(bool): Calculate scores for predicted results. + score_true_data(str, pandas.DataFrame, dict): Data with true values to calculate scores. If missed, target from filename used for true values. output(str): Output csv file path. no_features_in_result(bool) : Do not return feature columns in prediction result. False by default locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model @@ -146,7 +149,9 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are stored in rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, + threshold, locally, data, columns, predicted_at, output, no_features_in_result, + score, score_true_data ) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 3e998b5c..2a55df5e 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -21,9 +21,11 @@ def evaluate(self, run_id = None): def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None ): return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) - def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): + def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, + predicted_at=None, output=None, no_features_in_result=None, score=False, score_true_data=None): return AugerModel(self.ctx).predict( - model_id, filename, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + model_id, filename, threshold, locally, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): return AugerModel(self.ctx).actuals( diff --git a/a2ml/api/auger/config.py b/a2ml/api/auger/config.py index 5331a4f8..0a9e47a5 100644 --- a/a2ml/api/auger/config.py +++ b/a2ml/api/auger/config.py @@ -7,12 +7,12 @@ def set_data_set(self, name, source=None, validation=False): #TODO: add more providers later if validation: self.ctx.config.set('experiment/validation_dataset', name) - if self.ctx.use_auger_cloud(): + if self.ctx.use_auger_cloud() and 'azure' in self.ctx.get_providers(): self.ctx.config.set('experiment/validation_dataset', name, "azure") else: - print("set_data_set: %s"%self.ctx.use_auger_cloud()) + #print("set_data_set: %s"%self.ctx.use_auger_cloud()) self.ctx.config.set('dataset', name) - if self.ctx.use_auger_cloud(): + if self.ctx.use_auger_cloud() and 'azure' in self.ctx.get_providers(): self.ctx.config.set('dataset', name, "azure") self.ctx.config.write_all() diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index e3232201..42057f2b 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -49,7 +49,8 @@ def check_endpoint(self, props=None): return is_endpoint - def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): + def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, + no_features_in_result=None, score=False, score_true_data=None): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') @@ -73,8 +74,9 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at prediction_api = AugerPredictionApi(self.ctx, self, use_endpoint=self.check_endpoint(props)) prediction_properties = \ prediction_api.create(records, features, threshold=threshold, file_url=file_url, - predicted_at=predicted_at, no_features_in_result=no_features_in_result) - + predicted_at=predicted_at, no_features_in_result=no_features_in_result, + score=score, score_true_data=score_true_data) + return prediction_properties.get('result') def actual(self, records, features, actuals_at, actuals_path, actual_date_column): diff --git a/a2ml/api/auger/impl/cloud/prediction.py b/a2ml/api/auger/impl/cloud/prediction.py index c9c76704..81f50387 100644 --- a/a2ml/api/auger/impl/cloud/prediction.py +++ b/a2ml/api/auger/impl/cloud/prediction.py @@ -14,7 +14,8 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self.parent_id_name = "endpoint_id" self._set_api_request_path("AugerEndpointPredictionApi") - def create(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): + def create(self, records, features, threshold=None, file_url=None, predicted_at=None, + no_features_in_result=None, score=False, score_true_data=None): params = { 'records': records, 'features': features, @@ -35,5 +36,11 @@ def create(self, records, features, threshold=None, file_url=None, predicted_at= if no_features_in_result is not None: params['no_features_in_result'] = no_features_in_result - + + if score: + params['score'] = score + + if score_true_data: + params['score_true_data'] = score_true_data + return self._call_create(params, ['requested', 'running']) diff --git a/a2ml/api/auger/impl/cloud/rest_api.py b/a2ml/api/auger/impl/cloud/rest_api.py index 58dcb9a7..fd1f39ae 100644 --- a/a2ml/api/auger/impl/cloud/rest_api.py +++ b/a2ml/api/auger/impl/cloud/rest_api.py @@ -25,11 +25,18 @@ def call_ex(self, method, params={}): if params.get('id') and not method.startswith('create_'): oid = params['id'] del params['id'] - #print(method, oid, params) - return getattr(self.hub_client, method)(oid, **params) + print(method, oid, params) + res = getattr(self.hub_client, method)(oid, **params) else: - #print(method, params) - return getattr(self.hub_client, method)(**params) + if method == 'create_endpoint_prediction' or method == 'create_endpoint_actual': + print(method, params.keys()) + else: + print(method, params) + + res = getattr(self.hub_client, method)(**params) + + #print(res) + return res def call(self, method, params={}): result = self.call_ex(method, params) diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index c6afa0e7..eeee9960 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -30,11 +30,13 @@ def review(self, model_id): def undeploy(self, model_id, locally=False): return ModelUndeploy(self.ctx, self.project).execute(model_id, locally) - def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): + def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, + output=None, no_features_in_result=None, score=False, score_true_data=None): if locally: self.deploy(model_id, locally) - return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, + predicted_at, output, no_features_in_result, score, score_true_data) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): if locally: diff --git a/a2ml/api/auger/impl/mparts/predict.py b/a2ml/api/auger/impl/mparts/predict.py index 005de202..ec19a7a2 100644 --- a/a2ml/api/auger/impl/mparts/predict.py +++ b/a2ml/api/auger/impl/mparts/predict.py @@ -23,7 +23,8 @@ def __init__(self, ctx): self.ctx = ctx def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, - predicted_at=None, output=None, no_features_in_result=None): + predicted_at=None, output=None, no_features_in_result=None, + score=False, score_true_data=None): if filename and not (filename.startswith("http:") or filename.startswith("https:")) and\ not fsclient.is_s3_path(filename): self.ctx.log('Predicting on data in %s' % filename) @@ -31,11 +32,14 @@ def execute(self, filename, model_id, threshold=None, locally=False, data=None, if locally: if locally == "docker": - predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) else: - predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) else: - predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) return predicted @@ -98,7 +102,8 @@ def _check_model_project(self, pipeline_api): raise AugerException("Project name: %s in config.yml is different from model project name: %s. Please change name in config.yml."%( self.ctx.config.get('name'), model_project_name)) - def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, + output, no_features_in_result, score, score_true_data): records, features, file_url, is_pandas_df = self._process_input(filename, data, columns) temp_file = None ds_result = None @@ -107,7 +112,8 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic else: pipeline_api = AugerPipelineApi(self.ctx, None, model_id) predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, - predicted_at=predicted_at, no_features_in_result=no_features_in_result) + predicted_at=predicted_at, no_features_in_result=no_features_in_result, + score=score, score_true_data=score_true_data) try: ds_result = DataFrame.create_dataframe(predictions.get('signed_prediction_url'), @@ -137,7 +143,8 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic if temp_file: fsclient.remove_file(temp_file) - def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, + output, no_features_in_result, score, score_true_data): from auger_ml.model_exporter import ModelExporter is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) @@ -148,6 +155,11 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre if columns is not None: columns = list(columns) + if score and score_true_data is None: + options = fsclient.read_json_file(os.path.join(model_path, "options.json")) + ds = DataFrame.create_dataframe(filename_arg, data, [options['targetFeature']]) + score_true_data = ds.df + res, options = ModelExporter({}).predict_by_model_to_ds(model_path, path_to_predict=filename_arg, records=data, features=columns, threshold=threshold, no_features_in_result=no_features_in_result) @@ -158,16 +170,25 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre if isinstance(data, pd.DataFrame): ds_result.from_pandas = True - return ModelHelper.save_prediction(ds_result, + predictions = ModelHelper.save_prediction(ds_result, prediction_id = None, json_result=False, count_in_result=False, prediction_date=predicted_at, model_path=model_path, model_id=model_id, output=output) + if not score: + return predictions + + scores = ModelExporter({}).score_by_model(model_path, predictions=predictions, + test_path = score_true_data) + + return {'predicted': predictions, 'scores': scores} + # return ModelExporter({}).predict_by_model(model_path=model_path, # path_to_predict=filename_arg, records=data, features=columns, # threshold=threshold, prediction_date=predicted_at, # no_features_in_result=no_features_in_result) #, output=output) - def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, + output, no_features_in_result, score, score_true_data): model_deploy = ModelDeploy(self.ctx, None) is_model_loaded, model_path = model_deploy.verify_local_model(model_id, add_model_folder=False) if not is_model_loaded: @@ -180,7 +201,7 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co filename = os.path.join(self.ctx.config.get_path(), '.augerml', 'predict_data.csv') ds.saveToCsvFile(filename, compression=None) - predicted = self._docker_run_predict(filename, threshold, model_path) + predicted = self._docker_run_predict(filename, threshold, model_path, score, score_true_data) if not filename_arg: ds_result = DataFrame.create_dataframe(predicted) @@ -198,13 +219,14 @@ def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, co return predicted - def _docker_run_predict(self, filename, threshold, model_path): + def _docker_run_predict(self, filename, threshold, model_path, score, score_true_data): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) docker_tag = cluster_settings.get('kubernetes_stack') predict_file = os.path.basename(filename) data_path = os.path.abspath(os.path.dirname(filename)) model_path = os.path.abspath(model_path) + #TODO: support score, score_true_data call_args = "--verbose=True --path_to_predict=./model_data/%s %s" % \ (predict_file, "--threshold=%s" % str(threshold) if threshold else '') diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 84f8c13a..d9011329 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -21,10 +21,12 @@ def deploy(self, project, model_id, locally, review, name, algorithm, score, dat @error_handler @authenticated - @with_project(autocreate=False) - def predict(self, project, filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result): - predicted = Model(self.ctx, project).predict( - filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) + #@with_project(autocreate=False) + def predict(self, filename, model_id, threshold, locally, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data): + predicted = Model(self.ctx, project=None).predict( + filename, model_id, threshold, locally, data, columns, predicted_at, output, + no_features_in_result, score, score_true_data) if filename: self.ctx.log('Predictions stored in %s' % predicted) diff --git a/a2ml/api/utils/dataframe.py b/a2ml/api/utils/dataframe.py index caf7c45b..ca737052 100644 --- a/a2ml/api/utils/dataframe.py +++ b/a2ml/api/utils/dataframe.py @@ -50,6 +50,9 @@ def create_dataframe(data_path=None, records=None, features=None): elif records is not None and isinstance(records, pd.DataFrame): ds = DataFrame({}) ds.df = records + if features: + ds.df = ds.df[features] + ds.from_pandas = True else: ds = DataFrame({}) diff --git a/a2ml/cmdl/commands/cmd_deploy.py b/a2ml/cmdl/commands/cmd_deploy.py index cd050865..0920c958 100644 --- a/a2ml/cmdl/commands/cmd_deploy.py +++ b/a2ml/cmdl/commands/cmd_deploy.py @@ -4,10 +4,10 @@ @click.command('deploy', short_help='Deploy trained model.') [email protected]('model-id', required=False, type=click.STRING) @click.option('--provider', '-p', type=click.Choice(['auger','azure','external']), required=False, help='Cloud AutoML Provider.') [email protected]('model-id', required=False, type=click.STRING) [email protected]('--locally', is_flag=True, default=False, [email protected]('--locally', '-l', is_flag=True, default=False, help='Download and deploy trained model locally.') @click.option('--no-review', is_flag=True, default=False, help='Do not support model review based on actual data.') diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index 8fedc0a6..c9291fbc 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -28,15 +28,19 @@ def cmdl(ctx): @pass_context def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score, data_path): """Deploy trained model.""" - A2MLModel(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score, data_path=data_path) + A2MLModel(ctx, provider).deploy(model_id, locally=locally, review=not no_review, + name=name, algorithm=algorithm, score=score, data_path=data_path) @click.command('predict', short_help='Predict with deployed model.') [email protected]('model-id', required=True, type=click.STRING) @click.argument('filename', required=True, type=click.STRING) @click.option('--threshold', '-t', default=None, type=float, help='Threshold.') [email protected]('--model-id', '-m', type=click.STRING, required=True, - help='Deployed model id.') [email protected]('--locally', is_flag=True, default=False, [email protected]('--score', '-s', is_flag=True, default=False, + help='Calculate scores for predicted results.') [email protected]('--score_true_path', type=click.STRING, required=False, + help='Path to true values to calculate scores. If missed, target from filename used for true values.') [email protected]('--locally', '-l', is_flag=True, default=False, help='Predict locally using auger.ai.predict package.') @click.option('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @@ -50,12 +54,13 @@ def predict(ctx, provider, filename, model_id, threshold, locally, docker, outpu if docker: locally = "docker" - A2MLModel(ctx, provider).predict(filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) + A2MLModel(ctx, provider).predict(model_id, filename=filename, + threshold=threshold, score=score, score_true_data=score_true_path, + locally=locally, output=output) @click.command('actuals', short_help='Send actual values for deployed model. Needed for review and monitoring.') [email protected]('model-id', required=True, type=click.STRING) @click.argument('filename', required=True, type=click.STRING) [email protected]('--model-id', '-m', type=click.STRING, required=True, - help='Deployed model id.') @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') @click.option('--locally', is_flag=True, default=False, @@ -98,8 +103,7 @@ def undeploy(ctx, provider, model_id, locally): A2MLModel(ctx, provider).undeploy(model_id, locally) @click.command('delete_actuals', short_help='Delete files with actuals and predcitions locally or from specified provider(s).') [email protected]('--model-id', '-m', type=click.STRING, required=True, - help='Deployed model id.') [email protected]('model-id', required=True, type=click.STRING) @click.option('--with-predictions', is_flag=True, default=False, help='Remove predictions.') @click.option('--begin-date', '-b', type=click.STRING, required=False, diff --git a/a2ml/cmdl/commands/cmd_predict.py b/a2ml/cmdl/commands/cmd_predict.py index 526c31cb..b27c00aa 100644 --- a/a2ml/cmdl/commands/cmd_predict.py +++ b/a2ml/cmdl/commands/cmd_predict.py @@ -4,24 +4,28 @@ @click.command('predict', short_help='Predict with deployed model.') [email protected]('--provider', '-p', type=click.Choice(['auger','azure']), required=False, - help='Cloud AutoML Provider.') [email protected]('model-id', required=True, type=click.STRING) @click.argument('filename', required=True, type=click.STRING) @click.option('--threshold', '-t', default=None, type=float, help='Threshold.') [email protected]('--model-id', '-m', type=click.STRING, required=False, - help='Deployed model id.') [email protected]('--locally', is_flag=True, default=False, [email protected]('--score', '-s', is_flag=True, default=False, + help='Calculate scores for predicted results.') [email protected]('--score_true_path', type=click.STRING, required=False, + help='Path to true values to calculate scores. If missed, target from filename used for true values.') [email protected]('--locally', '-l', is_flag=True, default=False, help='Predict locally using auger.ai.predict package.') @click.option('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @click.option('--output', '-o', type=click.STRING, required=False, help='Output csv file path.') [email protected]('--provider', '-p', type=click.Choice(['auger','azure']), required=False, + help='Cloud AutoML Provider.') @pass_context def cmdl(ctx, provider, filename, model_id, threshold, locally, docker, output): """Predict with deployed model.""" ctx.setup_logger(format='') if docker: locally = "docker" - A2ML(ctx, provider).predict( - filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) + A2ML(ctx, provider).predict( model_id, filename=filename, + threshold=threshold, score=score, score_true_data=score_true_path, + locally=locally, output=output)

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-08-08T19:21:50

0.0

[]

augerai/a2ml

augerai__a2ml-582

4963565d278901b20ce8715959270a36d4c2d053

diff --git a/a2ml/api/roi/var_names_fetcher.py b/a2ml/api/roi/var_names_fetcher.py index 2bbeb87c..355c8418 100644 --- a/a2ml/api/roi/var_names_fetcher.py +++ b/a2ml/api/roi/var_names_fetcher.py @@ -36,6 +36,9 @@ def evaluate_unary_op_node(self, node): def evaluate_func_node(self, node): list(map(lambda node: self.evaluate(node), node.arg_nodes)) + def evaluate_tuple_node(self, node): + list(map(lambda node: self.evaluate(node), node.item_nodes)) + def evaluate_top_node(self, node): list(map(lambda n: self.evaluate(n), node.child_nodes()))

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-07-05T18:24:00

0.0

[]

augerai/a2ml

augerai__a2ml-578

2486bd45dfae8cf149e8dec88c18ba63b7f1141a

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index ac1a0a5d..cf42dc04 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.28' +__version__ = '1.0.29' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 32308b80..09fe4116 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -314,7 +314,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value. If missed - predict called automatically + * - predicted( or target): predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 9730d888..2be596f0 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -159,7 +159,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value. If missed - predict called automatically + * - predicted ( or target): predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index 46130f39..d641dc8c 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -169,8 +169,11 @@ def add_actuals( actuals_count = ds_actuals.count() ds_actuals.df.rename(columns={"actual": 'a2ml_actual'}, inplace=True) + if 'predicted' in ds_actuals.columns and not self.target_feature in ds_actuals.columns: + ds_actuals.df = ds_actuals.df.rename(columns={'predicted': self.target_feature}) + if provider is not None and (do_predict or not self.target_feature in ds_actuals.columns): - logging.info("Actual data missing predicted value column: %s. Call predict with features from actual data: %s"%(self.target_feature, ds_actuals.columns)) + logging.info("Actual data missing 'predicted' column and predicted value column: %s. Call predict with features from actual data: %s"%(self.target_feature, ds_actuals.columns)) self._do_predict(ctx, ds_actuals, provider) result = self._do_score_actual(ds_actuals.df)

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-07-02T19:47:17

0.0

[]

augerai/a2ml

augerai__a2ml-560

6cf1f1c1e471b7101fbf8f510e8ef3cec6bacc88

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index dce8b34b..fa7c0d49 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.19' +__version__ = '1.0.20' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index 0a727c3b..921f8892 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -221,7 +221,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, locally=False, provider=None): + threshold=None, output=None, no_features_in_result = None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -236,7 +236,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. output(str): Output csv file path. - locally(bool): Predicts using a local model if True, on the Provider Cloud if False. + no_features_in_result(bool) : Do not return feature columns in prediction result. False by default + locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. Returns: @@ -287,7 +288,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are returned as rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index da68a419..4df4de90 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -66,7 +66,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, - threshold=None, output=None, locally=False, provider=None): + threshold=None, output=None, no_features_in_result=None, locally=False, provider=None): """Predict results with new data against deployed model. Predictions are stored next to the file with data to be predicted on. The file name will be appended with suffix _predicted. Note: @@ -81,7 +81,8 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at predicted_at: Predict data date. Use for review of historical data. threshold(float): For classification models only. This will return class probabilities with response. output(str): Output csv file path. - locally(bool): Predicts using a local model if True, on the Provider Cloud if False. + no_features_in_result(bool) : Do not return feature columns in prediction result. False by default + locally(bool, str): Predicts using a local model with auger.ai.predict if True, on the Provider Cloud if False. If set to "docker", then docker image used to run the model provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider set in costructor or config. Returns: @@ -132,7 +133,7 @@ def predict(self, model_id, filename=None, data=None, columns=None, predicted_at # predictions are returned as rv[provider]['data']['predicted'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output) + return self.get_runner(locally, model_id, provider).execute_one_provider('predict', filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) @show_result def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False, provider=None): diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 5bc25d38..3e998b5c 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -21,9 +21,9 @@ def evaluate(self, run_id = None): def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None ): return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) - def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): return AugerModel(self.ctx).predict( - model_id, filename, threshold, locally, data, columns, predicted_at, output) + model_id, filename, threshold, locally, data, columns, predicted_at, output, no_features_in_result) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): return AugerModel(self.ctx).actuals( diff --git a/a2ml/api/auger/impl/cloud/experiment_session.py b/a2ml/api/auger/impl/cloud/experiment_session.py index 7b6a474d..2668f8fe 100644 --- a/a2ml/api/auger/impl/cloud/experiment_session.py +++ b/a2ml/api/auger/impl/cloud/experiment_session.py @@ -72,6 +72,8 @@ def get_leaderboard(self): '{0:.4f}'.format(item.get('score_value')) } review_metric = self.ctx.config.get('review/metric') + if review_metric == 'roi': + review_metric = 'mda' if review_metric: l_item[review_metric] = \ '{0:.4f}'.format(item.get('raw_data', {}).get('all_scores', {}).get(review_metric, 0.0)) diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index 8149b184..e3232201 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -49,7 +49,7 @@ def check_endpoint(self, props=None): return is_endpoint - def predict(self, records, features, threshold=None, file_url=None, predicted_at=None): + def predict(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): if self.object_id is None: raise AugerException('Please provide Auger Pipeline id') @@ -72,7 +72,9 @@ def predict(self, records, features, threshold=None, file_url=None, predicted_at prediction_api = AugerPredictionApi(self.ctx, self, use_endpoint=self.check_endpoint(props)) prediction_properties = \ - prediction_api.create(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at) + prediction_api.create(records, features, threshold=threshold, file_url=file_url, + predicted_at=predicted_at, no_features_in_result=no_features_in_result) + return prediction_properties.get('result') def actual(self, records, features, actuals_at, actuals_path, actual_date_column): diff --git a/a2ml/api/auger/impl/cloud/pipeline_file.py b/a2ml/api/auger/impl/cloud/pipeline_file.py index ddeee2ec..bb7e65db 100644 --- a/a2ml/api/auger/impl/cloud/pipeline_file.py +++ b/a2ml/api/auger/impl/cloud/pipeline_file.py @@ -31,7 +31,7 @@ def download(self, url, path_to_download, trial_id): return file_name def _get_status_name(self): - return 's3_model_path_status' + return 'signed_s3_model_path_status' def _log_status(self, status): if status is None: diff --git a/a2ml/api/auger/impl/cloud/prediction.py b/a2ml/api/auger/impl/cloud/prediction.py index 67f85dd5..c9c76704 100644 --- a/a2ml/api/auger/impl/cloud/prediction.py +++ b/a2ml/api/auger/impl/cloud/prediction.py @@ -14,7 +14,7 @@ def __init__(self, ctx, pipeline_api, use_endpoint=False): self.parent_id_name = "endpoint_id" self._set_api_request_path("AugerEndpointPredictionApi") - def create(self, records, features, threshold=None, file_url=None, predicted_at=None): + def create(self, records, features, threshold=None, file_url=None, predicted_at=None, no_features_in_result=None): params = { 'records': records, 'features': features, @@ -33,4 +33,7 @@ def create(self, records, features, threshold=None, file_url=None, predicted_at= if predicted_at: params['predicted_at'] = str(predicted_at) + if no_features_in_result is not None: + params['no_features_in_result'] = no_features_in_result + return self._call_create(params, ['requested', 'running']) diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 3ece907e..c6afa0e7 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -30,23 +30,21 @@ def review(self, model_id): def undeploy(self, model_id, locally=False): return ModelUndeploy(self.ctx, self.project).execute(model_id, locally) - def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): if locally: self.deploy(model_id, locally) - return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output) + return ModelPredict(self.ctx).execute(filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): if locally: - is_loaded, model_path, model_name = ModelDeploy(self.ctx, self.project).\ - verify_local_model(model_id) + is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) if not is_loaded: raise AugerException('Model should be deployed locally.') - model_path, model_existed = ModelPredict(self.ctx)._extract_model(model_name) params = { - 'model_path': os.path.join(model_path, "model"), + 'model_path': model_path, 'roi': { 'filter': str(self.ctx.config.get('review/roi/filter')), 'revenue': str(self.ctx.config.get('review/roi/revenue')), @@ -67,28 +65,22 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N def delete_actuals(self, model_id, with_predictions=False, begin_date=None, end_date=None, locally=False): if locally: - is_loaded, model_path, model_name = ModelDeploy(self.ctx, self.project).\ - verify_local_model(model_id) - + is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) if not is_loaded: raise AugerException('Model should be deployed locally.') - model_path, model_existed = ModelPredict(self.ctx)._extract_model(model_name) - return ModelReview({'model_path': os.path.join(model_path, "model")}).delete_actuals( + return ModelReview({'model_path': model_path}).delete_actuals( with_predictions=with_predictions, begin_date=begin_date, end_date=end_date) else: return ModelDeleteActual(self.ctx).execute(model_id, with_predictions, begin_date, end_date) def build_review_data(self, model_id, locally, output): if locally: - is_loaded, model_path, model_name = ModelDeploy(self.ctx, self.project).\ - verify_local_model(model_id) - + is_loaded, model_path = ModelDeploy(self.ctx, self.project).verify_local_model(model_id) if not is_loaded: raise AugerException('Model should be deployed locally.') - model_path, model_existed = ModelPredict(self.ctx)._extract_model(model_name) - return ModelReview({'model_path': os.path.join(model_path, "model")}).build_review_data( + return ModelReview({'model_path': model_path}).build_review_data( data_path=self.ctx.config.get("source"), output=output) else: raise Exception("Not Implemented.") diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index be68bfcc..9eeeaf2a 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -23,7 +23,7 @@ def __init__(self, ctx, project): def execute(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): if locally: - return self.deploy_model_locally(model_id, review, name, data_path) + return self.deploy_model_locally(model_id, review, name, data_path, locally) else: return self.deploy_model_in_cloud(model_id, review, name, algorithm, score, data_path) @@ -159,36 +159,65 @@ def deploy_model_in_cloud(self, model_id, review, name, algorithm, score, data_p return pipeline_properties.get('id') - def deploy_model_locally(self, model_id, review, name, data_path): - is_loaded, model_path, model_name = self.verify_local_model(model_id) + def deploy_model_locally(self, model_id, review, name, data_path, locally): + is_loaded, model_path = self.verify_local_model(model_id) #TODO: support review flag if not is_loaded: self.ctx.log('Downloading model %s' % model_id) self.project.start() + models_path = os.path.join(self.ctx.config.get_path(), 'models') pipeline_file_api = AugerPipelineFileApi(self.ctx, None) pipeline_file_properties = pipeline_file_api.create(model_id) downloaded_model_file = pipeline_file_api.download( pipeline_file_properties['signed_s3_model_path'], - model_path, model_id) + models_path, model_id) self.ctx.log('Downloaded model to %s' % downloaded_model_file) - self.ctx.log('Pulling docker image required to predict') - self._docker_pull_image() + if locally == 'docker': + self.ctx.log('Pulling docker image required to predict') + self._docker_pull_image() + else: + self.ctx.log('To run predict locally install a2ml[predict]') else: - self.ctx.log('Downloaded model is %s' % model_name) + self.ctx.log('Downloaded model is %s' % model_path) return model_id - def verify_local_model(self, model_id): - model_path = os.path.join(self.ctx.config.get_path(), 'models') - model_name = os.path.join(model_path, 'model-%s.zip' % model_id) - is_exists = fsclient.is_folder_exists(os.path.join(model_path,"model-%s"%model_id)) - if not is_exists: - is_exists = fsclient.is_file_exists(model_name) - return is_exists, model_path, model_name + def get_local_model_paths(self, model_id): + models_path = os.path.join(self.ctx.config.get_path(), 'models') + model_zip_path = os.path.join(models_path, 'model-%s.zip' % model_id) + model_path = os.path.join(models_path,"model-%s"%model_id) + + return model_path, model_zip_path + + def verify_local_model(self, model_id, add_model_folder=True): + model_path, model_zip_path = self.get_local_model_paths(model_id) + + is_exists = fsclient.is_folder_exists(model_path) + if not is_exists and fsclient.is_file_exists(model_zip_path): + self._extract_model(model_zip_path) + + if add_model_folder: + model_path = os.path.join(model_path, "model") + + is_exists = fsclient.is_folder_exists(model_path) + + return is_exists, model_path + + def _extract_model(self, model_name): + from zipfile import ZipFile + + model_path = os.path.splitext(model_name)[0] + model_existed = os.path.exists(model_path) + + if not model_existed: + with ZipFile(model_name, 'r') as zip_file: + zip_file.extractall(model_path) + + return model_path, model_existed def _docker_pull_image(self): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) diff --git a/a2ml/api/auger/impl/mparts/predict.py b/a2ml/api/auger/impl/mparts/predict.py index 510e4ce4..005de202 100644 --- a/a2ml/api/auger/impl/mparts/predict.py +++ b/a2ml/api/auger/impl/mparts/predict.py @@ -1,7 +1,6 @@ import os import shutil import subprocess -from zipfile import ZipFile import sys import pandas as pd @@ -23,16 +22,20 @@ def __init__(self, ctx): super(ModelPredict, self).__init__() self.ctx = ctx - def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def execute(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, + predicted_at=None, output=None, no_features_in_result=None): if filename and not (filename.startswith("http:") or filename.startswith("https:")) and\ not fsclient.is_s3_path(filename): self.ctx.log('Predicting on data in %s' % filename) filename = os.path.abspath(filename) if locally: - predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output) + if locally == "docker": + predicted = self._predict_locally_in_docker(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) + else: + predicted = self._predict_locally(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) else: - predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output) + predicted = self._predict_on_cloud(filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result) return predicted @@ -95,7 +98,7 @@ def _check_model_project(self, pipeline_api): raise AugerException("Project name: %s in config.yml is different from model project name: %s. Please change name in config.yml."%( self.ctx.config.get('name'), model_project_name)) - def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, output): + def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): records, features, file_url, is_pandas_df = self._process_input(filename, data, columns) temp_file = None ds_result = None @@ -103,7 +106,8 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic ds_result = DataFrame.create_dataframe(None, [], features+[self.ctx.config.get('target')]) else: pipeline_api = AugerPipelineApi(self.ctx, None, model_id) - predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, predicted_at=predicted_at) + predictions = pipeline_api.predict(records, features, threshold=threshold, file_url=file_url, + predicted_at=predicted_at, no_features_in_result=no_features_in_result) try: ds_result = DataFrame.create_dataframe(predictions.get('signed_prediction_url'), @@ -121,11 +125,10 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic ds_result.loaded_columns = columns ds_result.from_pandas = is_pandas_df - is_model_loaded, model_path_1, model_name = \ - ModelDeploy(self.ctx, None).verify_local_model(model_id) - model_path = None - if is_model_loaded: - model_path = os.path.join(model_path_1, "model-%s"%model_id, 'model') + # Save prediction in local model folder if exist + is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) + if not is_model_loaded: + model_path = None return ModelHelper.save_prediction(ds_result, prediction_id = None, json_result=False, count_in_result=False, prediction_date=predicted_at, @@ -134,17 +137,42 @@ def _predict_on_cloud(self, filename, model_id, threshold, data, columns, predic if temp_file: fsclient.remove_file(temp_file) - def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, output): - model_deploy = ModelDeploy(self.ctx, None) - is_model_loaded, model_path, model_name = \ - model_deploy.verify_local_model(model_id) + def _predict_locally(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + from auger_ml.model_exporter import ModelExporter + is_model_loaded, model_path = ModelDeploy(self.ctx, None).verify_local_model(model_id) if not is_model_loaded: raise AugerException('Model isn\'t loaded locally. ' 'Please use a2ml deploy command to download model.') - model_path, model_existed = self._extract_model(model_name) - model_options = fsclient.read_json_file(os.path.join(model_path, "model", "options.json")) + if columns is not None: + columns = list(columns) + + res, options = ModelExporter({}).predict_by_model_to_ds(model_path, + path_to_predict=filename_arg, records=data, features=columns, + threshold=threshold, no_features_in_result=no_features_in_result) + + ds_result = DataFrame({'data_path': None}) + ds_result.df = res.df + ds_result.loaded_columns = columns + if isinstance(data, pd.DataFrame): + ds_result.from_pandas = True + + return ModelHelper.save_prediction(ds_result, + prediction_id = None, json_result=False, count_in_result=False, prediction_date=predicted_at, + model_path=model_path, model_id=model_id, output=output) + + # return ModelExporter({}).predict_by_model(model_path=model_path, + # path_to_predict=filename_arg, records=data, features=columns, + # threshold=threshold, prediction_date=predicted_at, + # no_features_in_result=no_features_in_result) #, output=output) + + def _predict_locally_in_docker(self, filename_arg, model_id, threshold, data, columns, predicted_at, output, no_features_in_result): + model_deploy = ModelDeploy(self.ctx, None) + is_model_loaded, model_path = model_deploy.verify_local_model(model_id, add_model_folder=False) + if not is_model_loaded: + raise AugerException('Model isn\'t loaded locally. ' + 'Please use a2ml deploy command to download model.') filename = filename_arg if not filename: @@ -152,15 +180,7 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre filename = os.path.join(self.ctx.config.get_path(), '.augerml', 'predict_data.csv') ds.saveToCsvFile(filename, compression=None) - try: - predicted = \ - self._docker_run_predict(filename, threshold, model_path) - finally: - # clean up unzipped model - # if it wasn't unzipped before - if not model_existed: - fsclient.remove_folder(model_path) - model_path = None + predicted = self._docker_run_predict(filename, threshold, model_path) if not filename_arg: ds_result = DataFrame.create_dataframe(predicted) @@ -178,16 +198,6 @@ def _predict_locally(self, filename_arg, model_id, threshold, data, columns, pre return predicted - def _extract_model(self, model_name): - model_path = os.path.splitext(model_name)[0] - model_existed = os.path.exists(model_path) - - if not model_existed: - with ZipFile(model_name, 'r') as zip_file: - zip_file.extractall(model_path) - - return model_path, model_existed - def _docker_run_predict(self, filename, threshold, model_path): cluster_settings = AugerClusterApi.get_cluster_settings(self.ctx) docker_tag = cluster_settings.get('kubernetes_stack') diff --git a/a2ml/api/auger/impl/mparts/undeploy.py b/a2ml/api/auger/impl/mparts/undeploy.py index 1dc26592..c15e2af2 100644 --- a/a2ml/api/auger/impl/mparts/undeploy.py +++ b/a2ml/api/auger/impl/mparts/undeploy.py @@ -17,15 +17,11 @@ def __init__(self, ctx, project): def execute(self, model_id, locally=False): if locally: - is_loaded, model_path, model_name = \ - ModelDeploy(self.ctx, self.project).verify_local_model(model_id) - self.ctx.log("Undeploy model. Remove local model: %s" % model_name) + model_path, model_zip_path = ModelDeploy(self.ctx, self.project).get_local_model_paths(model_id) + self.ctx.log("Undeploy model. Remove local model: %s" % model_path) - if is_loaded: - fsclient.remove_file(model_name) - - model_folder = os.path.splitext(model_name)[0] - fsclient.remove_folder(model_folder) + fsclient.remove_file(model_zip_path) + fsclient.remove_folder(model_path) else: pipeline_api = AugerPipelineApi(self.ctx, None, model_id) if pipeline_api.check_endpoint(): diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 24bbd661..84f8c13a 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -22,9 +22,9 @@ def deploy(self, project, model_id, locally, review, name, algorithm, score, dat @error_handler @authenticated @with_project(autocreate=False) - def predict(self, project, filename, model_id, threshold, locally, data, columns, predicted_at, output): + def predict(self, project, filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result): predicted = Model(self.ctx, project).predict( - filename, model_id, threshold, locally, data, columns, predicted_at, output) + filename, model_id, threshold, locally, data, columns, predicted_at, output, no_features_in_result) if filename: self.ctx.log('Predictions stored in %s' % predicted) diff --git a/a2ml/api/azure/a2ml.py b/a2ml/api/azure/a2ml.py index 078a7c9b..ef265de8 100644 --- a/a2ml/api/azure/a2ml.py +++ b/a2ml/api/azure/a2ml.py @@ -26,12 +26,12 @@ def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None return AzureModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) - def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): + def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None, no_features_in_result=None): from a2ml.api.azure.model import AzureModel return AzureModel(self.ctx).predict( filename, model_id, threshold=threshold, locally=locally, data=data, columns=columns, - predicted_at=predicted_at, output=output) + predicted_at=predicted_at, output=output, no_features_in_result=no_features_in_result) def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=None, actual_date_column=None, locally=False): from a2ml.api.azure.model import AzureModel diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index 5f7fb49f..8b473826 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -239,8 +239,8 @@ def get_df(data): @error_handler @authenticated def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, - predicted_at=None, output=None, json_result=False, count_in_result=False, prediction_id=None - ): + predicted_at=None, output=None, json_result=False, count_in_result=False, prediction_id=None, + no_features_in_result=None): ds = DataFrame.create_dataframe(filename, data, columns) model_path = self.ctx.config.get_model_path(model_id) options = fsclient.read_json_file(os.path.join(model_path, "options.json")) diff --git a/a2ml/api/model_review/model_helper.py b/a2ml/api/model_review/model_helper.py index 355b49a1..d2244c38 100644 --- a/a2ml/api/model_review/model_helper.py +++ b/a2ml/api/model_review/model_helper.py @@ -100,7 +100,7 @@ def save_metric(metric_id, project_path, metric_name, metric_data): # @staticmethod # def _get_score_byname(scoring): - # from sklearn.metrics.scorer import get_scorer + # from sklearn.metrics import get_scorer # from sklearn.metrics import SCORERS # #TODO: below metrics does not directly map to sklearn: @@ -157,7 +157,7 @@ def save_metric(metric_id, project_path, metric_name, metric_data): @staticmethod def calculate_scores(options, y_test, X_test=None, estimator=None, y_pred=None, raise_main_score=True): - from sklearn.metrics.scorer import get_scorer + from sklearn.metrics import get_scorer from sklearn.model_selection._validation import _score from sklearn.metrics import confusion_matrix diff --git a/a2ml/api/model_review/scores/classification.py b/a2ml/api/model_review/scores/classification.py index 76493cc3..3acd890d 100644 --- a/a2ml/api/model_review/scores/classification.py +++ b/a2ml/api/model_review/scores/classification.py @@ -3,7 +3,7 @@ from sklearn.metrics import make_scorer, recall_score, average_precision_score, roc_auc_score from sklearn.metrics import confusion_matrix from sklearn.metrics import matthews_corrcoef as mcc -from sklearn.metrics.scorer import SCORERS +from sklearn.metrics import SCORERS def kappa(y_true, y_pred, weights=None, allow_off_by_one=False): diff --git a/a2ml/api/model_review/scores/regression.py b/a2ml/api/model_review/scores/regression.py index b8cdf0bb..704f4539 100644 --- a/a2ml/api/model_review/scores/regression.py +++ b/a2ml/api/model_review/scores/regression.py @@ -1,6 +1,6 @@ import numpy as np from sklearn.metrics import make_scorer, mean_squared_error, mean_squared_log_error, mean_absolute_error -from sklearn.metrics.scorer import SCORERS +from sklearn.metrics import SCORERS EPSILON = 1e-10 diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index a380e6ac..8fedc0a6 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -37,14 +37,19 @@ def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score, @click.option('--model-id', '-m', type=click.STRING, required=True, help='Deployed model id.') @click.option('--locally', is_flag=True, default=False, + help='Predict locally using auger.ai.predict package.') [email protected]('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') @click.option('--output', '-o', type=click.STRING, required=False, help='Output csv file path.') @pass_context -def predict(ctx, provider, filename, model_id, threshold, locally, output): +def predict(ctx, provider, filename, model_id, threshold, locally, docker, output): """Predict with deployed model.""" + if docker: + locally = "docker" + A2MLModel(ctx, provider).predict(filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) @click.command('actuals', short_help='Send actual values for deployed model. Needed for review and monitoring.') diff --git a/a2ml/cmdl/commands/cmd_predict.py b/a2ml/cmdl/commands/cmd_predict.py index 5665273d..526c31cb 100644 --- a/a2ml/cmdl/commands/cmd_predict.py +++ b/a2ml/cmdl/commands/cmd_predict.py @@ -12,12 +12,16 @@ @click.option('--model-id', '-m', type=click.STRING, required=False, help='Deployed model id.') @click.option('--locally', is_flag=True, default=False, + help='Predict locally using auger.ai.predict package.') [email protected]('--docker', is_flag=True, default=False, help='Predict locally using Docker image to run model.') @click.option('--output', '-o', type=click.STRING, required=False, help='Output csv file path.') @pass_context -def cmdl(ctx, provider, filename, model_id, threshold, locally, output): +def cmdl(ctx, provider, filename, model_id, threshold, locally, docker, output): """Predict with deployed model.""" ctx.setup_logger(format='') + if docker: + locally = "docker" A2ML(ctx, provider).predict( filename=filename, model_id=model_id, threshold=threshold, locally=locally, output=output) diff --git a/a2ml/tasks_queue/tasks_hub_api.py b/a2ml/tasks_queue/tasks_hub_api.py index 534adde5..5340fca5 100644 --- a/a2ml/tasks_queue/tasks_hub_api.py +++ b/a2ml/tasks_queue/tasks_hub_api.py @@ -531,7 +531,8 @@ def predict_by_model_task(params): count_in_result=params.get('count_in_result'), predicted_at=params.get('prediction_date'), prediction_id = params.get('prediction_id'), - locally = params.get('locally', False) + locally = params.get('locally', False), + no_features_in_result=params.get('no_features_in_result', False) ) _update_hub_objects(ctx, params.get('provider'), params) diff --git a/setup.py b/setup.py index cf55b2a8..1a73562d 100644 --- a/setup.py +++ b/setup.py @@ -28,9 +28,9 @@ def run(self): install_requires = [ - 'numpy<1.19.0,>=1.16.0', # version for azure - 'pandas>=0.22', # version for azure - 'joblib>=0.14.1', # version for azure + 'numpy==1.18.5', # version for azure + 'pandas==1.2.4', # version for azure + 'joblib==1.0.1', # version for azure 'ruamel.yaml>0.16.7', # version for azure 'pyarrow<2.0.0,>=0.17.0', # version for azure 'scipy==1.5.2', @@ -72,23 +72,30 @@ def run(self): 'redis', 's3fs>=0.4.0,<0.5.0', 'uvicorn', + 'scikit-learn==0.24.2' ], 'azure': [ - 'scikit-learn~=0.22.2', - 'xgboost<=0.90', + #'scikit-learn~=0.22.2', + #'xgboost<=0.90', # https://github.com/Azure/azure-sdk-for-python/issues/13871 #'azure-mgmt-resource==10.2.0', - 'azureml-sdk[automl]~=1.22.0' + #this needs to move to setup.azure.py and do not include default + 'azureml-sdk[automl]==1.29.0' ], 'google': [ 'google-cloud-automl' + ], + 'predict': [ + 'auger.ai.predict==1.0.72' ] } # Meta dependency groups. all_deps = [] for group_name in extras: - all_deps += extras[group_name] + if group_name != 'predict' and group_name != 'google' and group_name != 'azure': + all_deps += extras[group_name] + extras['all'] = all_deps

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-05-27T20:31:05

0.0

[]

augerai/a2ml

augerai__a2ml-547

84094892b7a1865b4d6a0f1d13b2af695b5f154b

diff --git a/a2ml/api/roi/grammar.bnf b/a2ml/api/roi/grammar.bnf index 73b56ba7..5701ae44 100644 --- a/a2ml/api/roi/grammar.bnf +++ b/a2ml/api/roi/grammar.bnf @@ -96,7 +96,7 @@ func_call_statement: | NAME '(' (expression (',' expression)*)? ')' top_expression: - | top_expression_type' NUMBER 'by' expression ['per' expression] ['where' expression] ['from' '(' top_expression ')'] + | top_expression_type' NUMBER 'by' shift_expr ['per' expression] ['where' expression] ['from' '(' top_expression ')'] top_expression_type: | 'top' diff --git a/a2ml/api/roi/parser.py b/a2ml/api/roi/parser.py index d2b798fa..d9c2a920 100644 --- a/a2ml/api/roi/parser.py +++ b/a2ml/api/roi/parser.py @@ -376,7 +376,7 @@ def top_expression(self): node.limit_node = self.const_node(Token.INT_CONST) self.eat(Token.BY) - node.order_node = self.expression() + node.order_node = self.shift_expr() if self.current_token.type == Token.PER: self.eat(Token.PER) diff --git a/a2ml/api/roi/validator.py b/a2ml/api/roi/validator.py index 03c2de16..b299bb03 100644 --- a/a2ml/api/roi/validator.py +++ b/a2ml/api/roi/validator.py @@ -1,6 +1,6 @@ from a2ml.api.roi.base_interpreter import BaseInterpreter from a2ml.api.roi.lexer import AstError, Lexer -from a2ml.api.roi.parser import Parser +from a2ml.api.roi.parser import Parser, TopNode class ValidationError(AstError): pass @@ -84,5 +84,8 @@ def evaluate_func_node(self, node): raise ValidationError(f"unknown function '{node.func_name}' at position {node.position()}") def evaluate_top_node(self, node): + if not isinstance(self.root, TopNode): + raise ValidationError(f"top or bottom expression cannot be used as an argument or operand") + return all(map(lambda n: self.evaluate(n), node.child_nodes()))

Check how and/or works in top expression e.g. ``` top 8 by P and $spread_pct < 0.5 from (top 1 by P per $symbol) $spread_pct < 0.5 and top 8 by P from (top 1 by P per $symbol) ```

2021-04-26T13:57:29

0.0

[]

augerai/a2ml

augerai__a2ml-545

0369a5aedaf72a0533c6e3890c1d668a8246fc23

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index cc08f086..9b076950 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.14' +__version__ = '1.0.15' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index aa213b05..0a727c3b 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -300,7 +300,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value + * - target: predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa @@ -310,7 +310,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N - Iris-virginica - Iris-virginica - It may also contain train features to retrain while Review(if target missed) and for distribution chart + It may also contain train features to predict(if target missed), retrain model while Review and for distribution chart This method support only one provider diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index c8764b19..da68a419 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -145,7 +145,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N :widths: 50 50 50 :header-rows: 1 - * - target: predicted value + * - target: predicted value. If missed - predict called automatically - actual - baseline_target: predicted value for baseline model (OPTIONAL) * - Iris-setosa @@ -155,7 +155,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N - Iris-virginica - Iris-virginica - It may also contain train features to retrain while Review(if target missed) and for distribution chart + It may also contain train features to predict(if target missed), retrain model while Review and for distribution chart This method support only one provider @@ -226,6 +226,12 @@ def review_alert(self, model_id, parameters = None, locally=False, provider=None * threshold (float) * sensitivity (int): The amount of time(in hours) this metric must be at or below the threshold to trigger the alert. + * threshold_policy (all_values/average_value/any_value) + + - all_values: Default value. Trigger an alert when all values in sensitivity below threshold + - average_value: Trigger an alert when average of values in sensitivity below threshold + - any_value: Trigger an alert when any value in sensitivity below threshold + * action (no/retrain/retrain_deploy) - no: no action should be executed diff --git a/a2ml/api/auger/impl/cloud/endpoint.py b/a2ml/api/auger/impl/cloud/endpoint.py index 27030333..e581af4a 100644 --- a/a2ml/api/auger/impl/cloud/endpoint.py +++ b/a2ml/api/auger/impl/cloud/endpoint.py @@ -14,8 +14,9 @@ def __init__(self, ctx, endpoint_api, endpoint_id=None): def create(self, pipeline_id, name): return self._call_create({'pipeline_id': pipeline_id, 'name': name},[]) - def update(self, name): - return self._call_update({ 'id': self.object_id, 'name': name}) + def update(self, params): + params['id'] = self.object_id + return self._call_update(params) def update_roi(self): roi_names = ['review/roi/filter', 'review/roi/investment', 'review/roi/revenue'] diff --git a/a2ml/api/auger/impl/cloud/experiment.py b/a2ml/api/auger/impl/cloud/experiment.py index 65194f98..45a6c715 100644 --- a/a2ml/api/auger/impl/cloud/experiment.py +++ b/a2ml/api/auger/impl/cloud/experiment.py @@ -96,8 +96,6 @@ def get_experiment_options(config, ): if config.get('experiment/search_space', None) is not None: options['search_space'] = config.get('experiment/search_space') - if config.get('review/metric'): - options['review_metric'] = config.get('review/metric') if config.get('review/alert/retrain_policy/type'): options['retrain_policy_type'] = config.get('review/alert/retrain_policy/type') if config.get('review/alert/retrain_policy/value'): diff --git a/a2ml/api/auger/impl/cloud/review_alert.py b/a2ml/api/auger/impl/cloud/review_alert.py index 30363621..821279ed 100644 --- a/a2ml/api/auger/impl/cloud/review_alert.py +++ b/a2ml/api/auger/impl/cloud/review_alert.py @@ -23,6 +23,7 @@ def create_update(self, parameters=None): 'active': parameters.get('active', config.get('review/alert/active')), 'kind': parameters.get('type', config.get('review/alert/type')), 'threshold': float(parameters.get('threshold', config.get('review/alert/threshold'))), + 'threshold_policy': parameters.get('threshold_policy', config.get('review/alert/threshold_policy')), 'sensitivity': int(parameters.get('sensitivity', config.get('review/alert/sensitivity'))), 'actions': parameters.get('action', config.get('review/alert/action')), 'notifications': parameters.get('notification', config.get('review/alert/notification')) diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index 61a32901..ddb7c98a 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -50,8 +50,12 @@ def create_update_review_alert(self, model_id, pipeline_properties=None, paramet endpoint_api = AugerEndpointApi(self.ctx, None, pipeline_properties['endpoint_pipelines'][0].get('endpoint_id')) + params = {'review_metric': self.ctx.config.get('review/metric')} if name and update_name: - endpoint_api.update(name) + params['name'] = name + + if params: + endpoint_api.update(params) session_id = endpoint_api.properties().get('primary_experiment_session_id') if session_id: diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index 43eae7b5..5f7fb49f 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -47,7 +47,6 @@ def deploy(self, model_id, locally, review, name=None, algorithm=None, score=Non 'scoreNames': [self.ctx.config.get('experiment/metric')], 'scoring': self.ctx.config.get('experiment/metric'), "score_name": self.ctx.config.get('experiment/metric'), - "review_metric": self.ctx.config.get('review/metric'), "originalFeatureColumns": model_features, "model_type": self.ctx.config.get("model_type") } diff --git a/a2ml/api/model_review/model_review.py b/a2ml/api/model_review/model_review.py index eec8cf41..916659c1 100644 --- a/a2ml/api/model_review/model_review.py +++ b/a2ml/api/model_review/model_review.py @@ -332,7 +332,6 @@ def score_model_performance_daily(self, date_from, date_to, extra_features=[]): res[str(curr_date)] = { 'scores': scores, 'score_name': self.options.get('score_name'), - 'review_metric': self.options.get('review_metric'), 'baseline_scores': baseline_score } diff --git a/a2ml/cmdl/template/config.yaml b/a2ml/cmdl/template/config.yaml index 566e9a90..2bfc8cc3 100644 --- a/a2ml/cmdl/template/config.yaml +++ b/a2ml/cmdl/template/config.yaml @@ -72,13 +72,19 @@ review: alert: # Activate/Deactivate Review Alert active: True - #model_accuracy - Decrease in Model Accuracy: the model accuracy threshold allowed before trigger is initiated. Default threshold: 0.7. Default sensitivity: 72 - #feature_average_range - Feature Average Out-Of-Range: Trigger an alert if average feature value during time period goes beyond the standard deviation range calculated during training period by the specified number of times or more. Default threshold: 1. Default sensitivity: 168 - #runtime_errors_burst - Burst Of Runtime Errors: Trigger an alert if runtime error count exceeds threshold. Default threshold: 5. Default sensitivity: 1 + # model_accuracy - Decrease in Model Accuracy: the model accuracy threshold allowed before trigger is initiated. Default threshold: 0.7. Default sensitivity: 72 + # feature_average_range - Feature Average Out-Of-Range: Trigger an alert if average feature value during time period goes beyond the standard deviation range calculated during training period by the specified number of times or more. Default threshold: 1. Default sensitivity: 168 + # runtime_errors_burst - Burst Of Runtime Errors: Trigger an alert if runtime error count exceeds threshold. Default threshold: 5. Default sensitivity: 1 type: model_accuracy threshold: 0.7 #The amount of time(in hours) this metric must be at or below the threshold to trigger the alert. sensitivity: 72 + + # all_values - Trigger an alert when all values in sensitivity below threshold. + # average_value - Trigger an alert when average of values in sensitivity below threshold. + # any_value - Trigger an alert when any value in sensitivity below threshold + threshold_policy: all_values + #no - no action should be executed #retrain - Use new predictions and actuals as test set to retrain the model. #retrain_deploy - Deploy retrained model and make it active model of this endpoint. diff --git a/docs/source/dev/configuration.rst b/docs/source/dev/configuration.rst index a8d2caee..c983ac6f 100644 --- a/docs/source/dev/configuration.rst +++ b/docs/source/dev/configuration.rst @@ -44,6 +44,7 @@ All Providers type: model_accuracy threshold: 0.7 sensitivity: 72 + threshold_policy: all_values action: retrain_deploy notification: user @@ -91,6 +92,13 @@ All Providers * **review.alert.threshold** Float * **review.alert.sensitivity** The amount of time(in hours) this metric must be at or below the threshold to trigger the alert. + * **review.alert.threshold_policy** + + * **Supported Review Alert threshold policies** + * **all_values** Trigger an alert when all values in sensitivity below threshold. + * **average_value** Trigger an alert when average of values in sensitivity below threshold. + * **any_value** Trigger an alert when any value in sensitivity below threshold. + * **review.alert.action** * **Supported Review Alert actions** diff --git a/docs/source/dev/mlram.rst b/docs/source/dev/mlram.rst index 941cf7d6..1da9d1cf 100644 --- a/docs/source/dev/mlram.rst +++ b/docs/source/dev/mlram.rst @@ -11,6 +11,8 @@ Auger model Monitored model =================== +See application example: https://github.com/augerai/mlram_apps + 1. Create A2ML application with external provider: .. code-block:: bash @@ -66,6 +68,9 @@ To review distribution chart , send training features with target and actuals: .. code-block:: python + # If call just after actuals, wait some time till server process the data + time.sleep(30) + ctx = Context() result = A2ML(ctx).review(model_id='external_model_id') if result['data']['status'] == 'retrain':

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-04-21T11:17:28

0.0

[]

augerai/a2ml

augerai__a2ml-536

71d7180c37deba8a27bee7b1c67f9dae2d39553e

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index b49950b1..0fb5dff1 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.11' +__version__ = '1.0.12' diff --git a/a2ml/api/auger/impl/cloud/base.py b/a2ml/api/auger/impl/cloud/base.py index 81c45ac1..5ea06a53 100644 --- a/a2ml/api/auger/impl/cloud/base.py +++ b/a2ml/api/auger/impl/cloud/base.py @@ -134,6 +134,7 @@ def _call_create(self, params=None, progress=None,has_return_object=True): if has_return_object: if object_properties: self.object_id = object_properties.get('id') + self.object_name = object_properties.get('name') #name can be changed by hub if progress: self.wait_for_status(progress) return self.properties()

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-03-22T06:21:43

0.0

[]

augerai/a2ml

augerai__a2ml-535

3088a5db6d2dc5fdb4c4489740411cf140b77a62

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index d521168a..b49950b1 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.10' +__version__ = '1.0.11' diff --git a/a2ml/api/a2ml.py b/a2ml/api/a2ml.py index c92955a7..3a14194f 100644 --- a/a2ml/api/a2ml.py +++ b/a2ml/api/a2ml.py @@ -175,7 +175,8 @@ def evaluate(self, run_id = None): return self.runner.execute('evaluate', run_id = run_id) @show_result - def deploy(self, model_id, locally=False, review=True, provider=None, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally=False, review=True, provider=None, + name=None, algorithm=None, score=None, data_path=None): """Deploy a model locally or to specified provider(s). Note: @@ -190,6 +191,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, name (str): Friendly name for the model. Used as name for Review Endpoint algorithm (str): Self-hosted model(external provider) algorithm name. score (float): Self-hosted model(external provider) score. + data_path (str): Data path to fit model when deploy. Return new deployed model-id Returns: :: @@ -214,7 +216,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, model_id = result['model_id'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', model_id, locally, review, name, algorithm, score) + return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', + model_id, locally, review, name, algorithm, score, data_path) @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index 83db12dd..b5400042 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -25,7 +25,8 @@ def __init__(self, ctx, provider=None): self.local_runner = lambda: self.build_runner(ctx, provider, force_local=True) @show_result - def deploy(self, model_id, locally=False, review=True, provider=None, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally=False, review=True, provider=None, + name=None, algorithm=None, score=None, data_path=None): """Deploy a model locally or to specified provider(s). Args: @@ -36,6 +37,7 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, name (str): Friendly name for the model. Used as name for Review Endpoint algorithm (str): Self-hosted model(external provider) algorithm name. score (float): Self-hosted model(external provider) score. + data_path (str): Data path to fit model when deploy. Return new deployed model-id Returns: :: @@ -59,7 +61,8 @@ def deploy(self, model_id, locally=False, review=True, provider=None, name=None, model_id = result['model_id'] """ - return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', model_id, locally, review, name, algorithm, score) + return self.get_runner(locally, model_id, provider).execute_one_provider('deploy', + model_id, locally, review, name, algorithm, score, data_path) @show_result def predict(self, model_id, filename=None, data=None, columns=None, predicted_at=None, diff --git a/a2ml/api/auger/a2ml.py b/a2ml/api/auger/a2ml.py index 015fff4e..5bc25d38 100644 --- a/a2ml/api/auger/a2ml.py +++ b/a2ml/api/auger/a2ml.py @@ -18,8 +18,8 @@ def train(self): def evaluate(self, run_id = None): return AugerExperiment(self.ctx).leaderboard(run_id) - def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): - return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score) + def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None ): + return AugerModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) def predict(self, model_id, filename, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): return AugerModel(self.ctx).predict( diff --git a/a2ml/api/auger/impl/cloud/pipeline.py b/a2ml/api/auger/impl/cloud/pipeline.py index 7e289e20..8149b184 100644 --- a/a2ml/api/auger/impl/cloud/pipeline.py +++ b/a2ml/api/auger/impl/cloud/pipeline.py @@ -11,8 +11,9 @@ def __init__(self, ctx, experiment_api, pipeline_id=None): super(AugerPipelineApi, self).__init__( ctx, experiment_api, None, pipeline_id) - def create(self, trial_id, review=True, name=None): - return self._call_create({'trial_id': trial_id, 'is_review_model_enabled' : review, 'name': name}, + def create(self, trial_id, review=True, name=None, refit_data_path=None): + return self._call_create({'trial_id': trial_id, 'is_review_model_enabled' : review, 'name': name, + 'refit_data_path': refit_data_path}, ['creating_files', 'packaging', 'deploying']) def create_external(self, review, name, project_id, algorithm, score): diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index 19ec06c7..3ece907e 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -18,8 +18,8 @@ def __init__(self, ctx, project): self.project = project self.ctx = ctx - def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): - return ModelDeploy(self.ctx, self.project).execute(model_id, locally, review, name, algorithm, score) + def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): + return ModelDeploy(self.ctx, self.project).execute(model_id, locally, review, name, algorithm, score, data_path) def review_alert(self, model_id, parameters, name): return ModelDeploy(self.ctx, self.project).create_update_review_alert(model_id, None, parameters, name) diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index 391cf465..90d020eb 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -21,11 +21,11 @@ def __init__(self, ctx, project): self.project = project self.ctx = ctx - def execute(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): + def execute(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): if locally: - return self.deploy_model_locally(model_id, review, name) + return self.deploy_model_locally(model_id, review, name, data_path) else: - return self.deploy_model_in_cloud(model_id, review, name, algorithm, score) + return self.deploy_model_in_cloud(model_id, review, name, algorithm, score, data_path) def create_update_review_alert(self, model_id, pipeline_properties=None, parameters=None, name=None): if not self.ctx.config.get('review'): @@ -121,7 +121,9 @@ def review(self, model_id): } return result - def deploy_model_in_cloud(self, model_id, review, name, algorithm, score): + def deploy_model_in_cloud(self, model_id, review, name, algorithm, score, data_path): + from .predict import ModelPredict + self.ctx.log('Deploying model %s' % model_id) if self.ctx.is_external_provider(): @@ -129,8 +131,12 @@ def deploy_model_in_cloud(self, model_id, review, name, algorithm, score): self.ctx, None).create_external(review, name, self.project.object_id, algorithm, score) else: self.project.start() + data_url = None + if data_path: + _, _, data_url, _ = ModelPredict(self.ctx)._process_input(data_path, None, None) + pipeline_properties = AugerPipelineApi( - self.ctx, None).create(model_id, review, name) + self.ctx, None).create(model_id, review, name, data_url) if pipeline_properties.get('status') == 'ready': if review: @@ -144,7 +150,7 @@ def deploy_model_in_cloud(self, model_id, review, name, algorithm, score): return pipeline_properties.get('id') - def deploy_model_locally(self, model_id, review, name): + def deploy_model_locally(self, model_id, review, name, data_path): is_loaded, model_path, model_name = self.verify_local_model(model_id) #TODO: support review flag if not is_loaded: diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index b901ee1c..24bbd661 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -15,8 +15,8 @@ def __init__(self, ctx): @error_handler @authenticated @with_project(autocreate=False) - def deploy(self, project, model_id, locally, review, name, algorithm, score): - model_id = Model(self.ctx, project).deploy(model_id, locally, review, name, algorithm, score) + def deploy(self, project, model_id, locally, review, name, algorithm, score, data_path): + model_id = Model(self.ctx, project).deploy(model_id, locally, review, name, algorithm, score, data_path) return {'model_id': model_id} @error_handler diff --git a/a2ml/api/azure/a2ml.py b/a2ml/api/azure/a2ml.py index a2647eee..078a7c9b 100644 --- a/a2ml/api/azure/a2ml.py +++ b/a2ml/api/azure/a2ml.py @@ -21,10 +21,10 @@ def evaluate(self, run_id = None): return AzureExperiment(self.ctx).leaderboard(run_id) - def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None, data_path=None): from a2ml.api.azure.model import AzureModel - return AzureModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score) + return AzureModel(self.ctx).deploy(model_id, locally, review, name, algorithm, score, data_path) def predict(self, filename, model_id, threshold=None, locally=False, data=None, columns=None, predicted_at=None, output=None): from a2ml.api.azure.model import AzureModel diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index 83195e43..43eae7b5 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -20,7 +20,7 @@ def __init__(self, ctx): @error_handler @authenticated - def deploy(self, model_id, locally, review, name=None, algorithm=None, score=None): + def deploy(self, model_id, locally, review, name=None, algorithm=None, score=None, data_path=None): if locally: is_loaded, model_path = self.verify_local_model(model_id) if is_loaded: diff --git a/a2ml/cmdl/commands/cmd_deploy.py b/a2ml/cmdl/commands/cmd_deploy.py index 474b1dae..0793b065 100644 --- a/a2ml/cmdl/commands/cmd_deploy.py +++ b/a2ml/cmdl/commands/cmd_deploy.py @@ -17,8 +17,10 @@ help='Self-hosted model(external provider) algorithm name.') @click.option('--score', '-s', required=False, type=float, help='Self-hosted model(external provider) score.') [email protected]('--data-path', '-d', type=click.STRING, required=False, + help='Data path to fit model when deploy. Return new deployed model-id') @pass_context -def cmdl(ctx, provider, model_id, locally, no_review, name, algorithm, score): +def cmdl(ctx, provider, model_id, locally, no_review, name, algorithm, score, data_path): """Deploy trained model.""" ctx.setup_logger(format='') - A2ML(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score) + A2ML(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score, data_path=data_path) diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index ccc5bacf..e6b187b4 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -23,10 +23,12 @@ def cmdl(ctx): help='Self-hosted model(external provider) algorithm name.') @click.option('--score', '-s', required=False, type=float, help='Self-hosted model(external provider) score.') [email protected]('--data-path', '-d', type=click.STRING, required=False, + help='Data path to fit model when deploy. Return new deployed model-id') @pass_context -def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score): +def deploy(ctx, provider, model_id, locally, no_review, name, algorithm, score, data_path): """Deploy trained model.""" - A2MLModel(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score) + A2MLModel(ctx, provider).deploy(model_id, locally, not no_review, name=name, algorithm=algorithm, score=score, data_path=data_path) @click.command('predict', short_help='Predict with deployed model.') @click.argument('filename', required=True, type=click.STRING)

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-03-09T17:48:33

0.0

[]

augerai/a2ml

augerai__a2ml-534

2e580917f270c6ed64528af1e34c08626d5a3fc0

diff --git a/a2ml/__init__.py b/a2ml/__init__.py index a67a991b..d521168a 100644 --- a/a2ml/__init__.py +++ b/a2ml/__init__.py @@ -1,1 +1,1 @@ -__version__ = '1.0.09' +__version__ = '1.0.10' diff --git a/a2ml/api/a2ml_model.py b/a2ml/api/a2ml_model.py index ed58746d..83db12dd 100644 --- a/a2ml/api/a2ml_model.py +++ b/a2ml/api/a2ml_model.py @@ -204,7 +204,7 @@ def actuals(self, model_id, filename=None, data=None, columns=None, actuals_at=N return self.get_runner(locally, model_id, provider).execute_one_provider('actuals', model_id, filename, data, columns, actuals_at, actual_date_column, locally) @show_result - def review_alert(self, model_id, parameters = None, locally=False, provider=None): + def review_alert(self, model_id, parameters = None, locally=False, provider=None, name=None): """Update Review parameters. Args: @@ -230,6 +230,7 @@ def review_alert(self, model_id, parameters = None, locally=False, provider=None locally(bool): Process review locally. provider (str): The automl provider you wish to run. For example 'auger'. The default is None - use provider defined by model_id or set in costructor. + name (str): Friendly name for the model. Used as name for Review Endpoint Returns: :: @@ -244,7 +245,7 @@ def review_alert(self, model_id, parameters = None, locally=False, provider=None ctx = Context() model = A2MLModel(ctx).review_alert(model_id='D881079E1ED14FB') """ - return self.get_runner(locally, model_id, provider).execute_one_provider('review_alert', model_id, parameters) + return self.get_runner(locally, model_id, provider).execute_one_provider('review_alert', model_id, parameters, name) @show_result def review(self, model_id, locally=False, provider=None): diff --git a/a2ml/api/auger/impl/cloud/endpoint.py b/a2ml/api/auger/impl/cloud/endpoint.py index eed9eb81..27030333 100644 --- a/a2ml/api/auger/impl/cloud/endpoint.py +++ b/a2ml/api/auger/impl/cloud/endpoint.py @@ -14,6 +14,9 @@ def __init__(self, ctx, endpoint_api, endpoint_id=None): def create(self, pipeline_id, name): return self._call_create({'pipeline_id': pipeline_id, 'name': name},[]) + def update(self, name): + return self._call_update({ 'id': self.object_id, 'name': name}) + def update_roi(self): roi_names = ['review/roi/filter', 'review/roi/investment', 'review/roi/revenue'] roi_values = [] diff --git a/a2ml/api/auger/impl/model.py b/a2ml/api/auger/impl/model.py index aedc93d4..19ec06c7 100644 --- a/a2ml/api/auger/impl/model.py +++ b/a2ml/api/auger/impl/model.py @@ -21,8 +21,8 @@ def __init__(self, ctx, project): def deploy(self, model_id, locally=False, review=True, name=None, algorithm=None, score=None): return ModelDeploy(self.ctx, self.project).execute(model_id, locally, review, name, algorithm, score) - def review_alert(self, model_id, parameters): - return ModelDeploy(self.ctx, self.project).create_update_review_alert(model_id, None, parameters) + def review_alert(self, model_id, parameters, name): + return ModelDeploy(self.ctx, self.project).create_update_review_alert(model_id, None, parameters, name) def review(self, model_id): return ModelDeploy(self.ctx, self.project).review(model_id) diff --git a/a2ml/api/auger/impl/mparts/deploy.py b/a2ml/api/auger/impl/mparts/deploy.py index 1023a93b..391cf465 100644 --- a/a2ml/api/auger/impl/mparts/deploy.py +++ b/a2ml/api/auger/impl/mparts/deploy.py @@ -34,7 +34,8 @@ def create_update_review_alert(self, model_id, pipeline_properties=None, paramet if not pipeline_properties: pipeline_properties = AugerPipelineApi(self.ctx, None, model_id).properties() - endpoint_api = None + endpoint_api = None + update_name = True if not pipeline_properties.get('endpoint_pipelines'): self.ctx.log('Creating review endpoint ...') endpoint_api = AugerEndpointApi(self.ctx, None) @@ -42,12 +43,16 @@ def create_update_review_alert(self, model_id, pipeline_properties=None, paramet name = fsclient.get_path_base_name(self.ctx.config.get('source')) endpoint_properties = endpoint_api.create(pipeline_properties.get('id'), name) pipeline_properties['endpoint_pipelines'] = [endpoint_properties.get('id')] + update_name = False if pipeline_properties.get('endpoint_pipelines'): if endpoint_api is None: endpoint_api = AugerEndpointApi(self.ctx, None, pipeline_properties['endpoint_pipelines'][0].get('endpoint_id')) + if name and update_name: + endpoint_api.update(name) + session_id = endpoint_api.properties().get('primary_experiment_session_id') if session_id: AugerExperimentSessionApi(self.ctx, None, None, session_id).update_settings() diff --git a/a2ml/api/auger/model.py b/a2ml/api/auger/model.py index 516c5c0f..b901ee1c 100644 --- a/a2ml/api/auger/model.py +++ b/a2ml/api/auger/model.py @@ -46,8 +46,8 @@ def delete_actuals(self, project, model_id, with_predictions=False, begin_date=N @error_handler @authenticated @with_project(autocreate=False) - def review_alert(self, project, model_id, parameters): - return Model(self.ctx, project).review_alert(model_id, parameters) + def review_alert(self, project, model_id, parameters, name): + return Model(self.ctx, project).review_alert(model_id, parameters, name) @error_handler @authenticated diff --git a/a2ml/api/azure/model.py b/a2ml/api/azure/model.py index e4342fd9..83195e43 100644 --- a/a2ml/api/azure/model.py +++ b/a2ml/api/azure/model.py @@ -554,7 +554,7 @@ def undeploy(self, model_id, locally): @error_handler @authenticated - def review_alert(self, model_id, parameters): + def review_alert(self, model_id, parameters, name): raise AzureException("Not Implemented. Set use_auger_cloud: True in config.yml") @error_handler diff --git a/a2ml/api/model_review/model_helper.py b/a2ml/api/model_review/model_helper.py index 64ceeb18..6a0c3541 100644 --- a/a2ml/api/model_review/model_helper.py +++ b/a2ml/api/model_review/model_helper.py @@ -4,7 +4,7 @@ import numpy as np import json -from a2ml.api.utils import get_uid, get_uid4, fsclient, remove_dups_from_list +from a2ml.api.utils import get_uid, get_uid4, fsclient, remove_dups_from_list, sort_arrays from a2ml.api.utils.dataframe import DataFrame @@ -184,6 +184,12 @@ def calculate_scores(options, y_test, X_test=None, estimator=None, y_pred=None, else: logging.error("calculate_scores: no scaling found for target fold group: %s"%options['fold_group']) + if options.get("score_top_count"): + if y_pred is None: + y_pred = estimator.predict(X_test) + + y_pred, y_test = sort_arrays(y_pred, y_test, options.get("score_top_count")) + all_scores = {} if y_pred is not None: if options.get('binaryClassification'): diff --git a/a2ml/api/utils/__init__.py b/a2ml/api/utils/__init__.py index 876bbf2e..f39f229e 100644 --- a/a2ml/api/utils/__init__.py +++ b/a2ml/api/utils/__init__.py @@ -249,4 +249,13 @@ def retry_helper(func, retry_errors=[], num_try=10, delay=10, ctx=None): time.sleep(delay*nTry) else: raise - + +def sort_arrays(ar1, ar2, top_n=None, desc=True): + p = ar1.argsort() + if desc: + p = p[::-1] + if top_n: + p = p[:top_n] + + return ar1[p], ar2[p] + diff --git a/a2ml/cmdl/commands/cmd_model.py b/a2ml/cmdl/commands/cmd_model.py index d2275b79..ccc5bacf 100644 --- a/a2ml/cmdl/commands/cmd_model.py +++ b/a2ml/cmdl/commands/cmd_model.py @@ -62,10 +62,12 @@ def actuals(ctx, provider, filename, model_id, locally): @click.argument('model-id', required=True, type=click.STRING) @click.option('--provider', '-p', type=click.Choice(['auger','azure']), required=False, help='Cloud AutoML Provider.') [email protected]('--name', '-n', required=False, type=click.STRING, + help='Model friendly name.Used as name for Review Endpoint') @pass_context -def review_alert(ctx, provider, model_id): +def review_alert(ctx, provider, model_id, name): """Predict with deployed model.""" - A2MLModel(ctx, provider).review_alert(model_id) + A2MLModel(ctx, provider).review_alert(model_id, name=name) @click.command('review', short_help='Review information about deployed model.') @click.argument('model-id', required=True, type=click.STRING)

WIP: Move api to auger.ai repo Moving all aunderlying auger api code to auger.ai repo

2021-03-03T11:26:13

0.0

[]