arthurai.explainability.arthur_explainer.EmptyLimeExplanation

class arthurai.explainability.arthur_explainer.EmptyLimeExplanation(class_int_labels, num_features)

Bases: object

Class for representing an all-NaN LIME Explanation. The LIME Explainers return full-featured lime.explanation.Explanation objects. We only use these to fetch the feature importance scores for each predicted class by calling as_map(). This class implements that as_map() method, and returns all-NaN feature importance scores.

Methods

as_map

rtype Dict[int, List[Tuple[int, float]]]

as_map()

Return type

Dict[int, List[Tuple[int, float]]]

Returns

A map from each predicted class integer label to a list of NaNs