eli5 0.0.5

Debug machine learning classifiers and explain their predictions

ELI5 is a Python package which helps to debug machine learning classifiers and explain their predictions.

Currently it allows to:

• explain weights and predictions of scikit-learn linear classifiers and regressors;

• explain weights of scikit-learn decision trees and tree-based ensemble classifiers (via feature importances);

• debug scikit-learn pipelines which contain HashingVectorizer, by undoing hashing;

• explain predictions of any black-box classifier using LIME ( http://arxiv.org/abs/1602.04938 ) algorithm.

TODO:

• IPython and HTML support

• https://github.com/TeamHG-Memex/sklearn-crfsuite and https://github.com/tpeng/python-crfsuite

• https://github.com/scikit-learn-contrib/polylearn

• https://github.com/scikit-learn-contrib/lightning

• fasttext (?)

• xgboost (?)

• eli5.lime improvements

• image input

• built-in support for non-text data in eli5.lime

• tensorflow, theano, lasagne, keras

• Naive Bayes from scikit-learn (see https://github.com/scikit-learn/scikit-learn/issues/2237)

• Reinforcement Learning support

• explain predictions of decision trees and treee-based ensembles

License is MIT.

Check docs for more (sorry, also TODO).

Changelog

0.0.5 (2016-09-27)

• HashingVectorizer support in explain_prediction;

• add an option to pass coefficient scaling array; it is useful if you want to compare coefficients for features which scale or sign is different in the input;

• bug fix: classifier weights are no longer changed by eli5 functions.

0.0.4 (2016-09-24)

• eli5.sklearn.InvertableHashingVectorizer and eli5.sklearn.FeatureUnhasher allow to recover feature names for pipelines which use HashingVectorizer or FeatureHasher;

• added support for scikit-learn linear regression models (ElasticNet, Lars, Lasso, LinearRegression, LinearSVR, Ridge, SGDRegressor);

• doc and vec arguments are swapped in explain_prediction function; vec can now be omitted if an example is already vectorized;

• fixed issue with dense feature vectors;

• all class_names arguments are renamed to target_names;

• feature name guessing is fixed for scikit-learn ensemble estimators;

• testing improvements.

0.0.3 (2016-09-21)

• support any black-box classifier using LIME (http://arxiv.org/abs/1602.04938) algorithm; text data support is built-in;

• “vectorized” argument for sklearn.explain_prediction; it allows to pass example which is already vectorized;

• allow to pass feature_names explicitly;

• support classifiers without get_feature_names method using auto-generated feature names.

0.0.2 (2016-09-19)

• ‘top’ argument of explain_prediction can be a tuple (num_positive, num_negative);

• classifier name is no longer printed by default;

• added eli5.sklearn.explain_prediction to explain individual examples;

• fixed numpy warning.

0.0.1 (2016-09-15)

Pre-release.