API

This is the full API documentation of the sharp package.

sharp.ShaRP([qoi, target_function, measure, ...])

Explains the contributions of features to different aspects of a ranked outcome, based on Shapley values.

sharp.qoi

Quantities of interest.

qoi.DiffQoI([target_function, X])	A general QoI, suitable for models/methods that output label predictions or scores.
qoi.FlipQoI([target_function, X])	Implements equation 4 from [R152a90246262-1].
qoi.LikelihoodQoI([target_function, X])	Implements equation 3 from [R383c5c9d41a2-1].
qoi.RankQoI([target_function, X])	Rank specific QoI.
qoi.RankScoreQoI([target_function, X])	A general, ranking-oriented QoI, similar to DiffQoI.
qoi.TopKQoI([target_function, top_k, X])	Rank-specific QoI.

qoi.get_qoi(qoi)	Get a quantity of interest from string.
qoi.get_qoi_names()	Get the names of all available quantities of interest.

sharp.visualization

Visualization functions based on matplotlib.

visualization.waterfall(xai, idx[, ...])

sharp.utils

Utility functions used to check variable types, convert scores to rankings, etc.

utils.check_feature_names(X)	Retrieve feature names from X.
utils.check_inputs(X[, y])	Converts X and y inputs to numpy arrays.
utils.check_measure(measure)	If None, return a default function.
utils.check_qoi(qoi[, target_function, X])	If None, return a default function.
utils.scores_to_ordering(y[, direction])	Converts an array with scores to a ranking.