constants Package

Contains classes defining constants used in interpretability in Azure Machine Learning.

For more information about interpretability, see Interpretability: model explanations in automated machine learning.

Classes

Attributes	Provide constants for attributes.
BackCompat	Provide constants necessary for supporting old versions of our product.
DNNFramework	Provide DNN framework constants.
Defaults	Provide constants for default values to explain methods.
Dynamic	Provide constants for dynamically generated classes.
ExplainParams	Provide constants for interpret community (init, explain_local and explain_global) parameters.
ExplainType	Provide constants for model and explainer type information, useful for visualization.
ExplanationParams	Provide constants for explanation parameters.
History	Provide constants related to uploading assets to run history.
IO	Provide file input and output related constants.
LightGBMParams	Provide constants for LightGBM.
LightGBMSerializationConstants	Provide internal class that defines fields used for MimicExplainer serialization.
LoggingNamespace	Provide logging namespace related constants.
MimicSerializationConstants	Provide internal class that defines fields used for MimicExplainer serialization.
RunPropertiesAndTags	Provide constants for tracking tags and properties set on the Run object.
SKLearn	Provide scikit-learn related constants.
Scoring	Provide constants for scoring time explainers.
Spacy	Provide spaCy related constants.
Tensorflow	Provide TensorFlow and TensorBoard related constants.

Enums

ExplainableModelType	Provide constants for the explainable model type.
ModelTask	Provide model task constants. Can be 'classification', 'regression', or 'unknown'. By default the model domain is inferred if 'unknown', but this can be overridden if you specify 'classification' or 'regression'.
ShapValuesOutput	Provide constants for the SHAP values output from the explainer. Can be 'default', 'probability' or 'teacher_probability'. If 'teacher_probability' is specified, we use the probabilities from the teacher model.