Generate an AntiNex AI Request¶
This method will use the environment variables from the consts.py:
generate_ai_request Method¶
-
antinex_client.generate_ai_request.
generate_ai_request
(predict_rows, req_dict=None, req_file='/opt/antinex/client/examples/predict-rows-scaler-full-django.json', features=[], ignore_features=[], sort_values=[], ml_type='classification', use_model_name='Full-Django-AntiNex-Simple-Scaler-DNN', predict_feature='label_value', seed=42, test_size=0.2, batch_size=32, epochs=15, num_splits=3, loss='binary_crossentropy', optimizer='adam', metrics=[], histories=[], filter_features_dict={}, filter_features=[], convert_enabled=True, convert_to_type='float', include_failed_conversions=False, value_for_missing='-1.0', version='1', publish_to_core=True, check_missing_predict_feature=True, debug=False)[source]¶ Parameters: - predict_rows – list of predict rows to build into the request
- req_dict – request dictionary to update - for long-running clients
- req_file – file holding a request dict to update - one-off tests
- features – features to process in the data
- ignore_features – features to ignore in the data (non-numerics)
- sort_values – optional - order rows for scaler normalization
- ml_type – machine learning type - classification/regression
- use_model_name – use a pre-trained model by name
- predict_feature – predict the values of this feature
- seed – seed for randomness reproducability
- test_size – split train/test data
- batch_size – batch size for processing
- epochs – test epochs
- num_splits – test splits for cross validation
- loss – loss function
- optimizer – optimizer
- metrics – metrics to apply
- histories – historical values to test
- filter_features_dict – dictionary of features to use
- filter_features – list of features to use
- convert_to_type – convert predict_row values to scaler-ready values
- include_failed_conversions – should the predict rows include fails
- value_for_missing – set this value to any columns that are missing
- version – version of the API request
- publish_to_core – want to publish it to the core or the worker
- debug – log debug messages