Identifying the Most Explainable Classifier

TL;DR

This paper introduces pointwise coverage to quantify classifier explainability and proves that binary linear classifiers are uniquely the most explainable under this measure.

Contribution

It defines a new explainability measure called pointwise coverage and characterizes binary linear classifiers as the most explainable classifiers according to this measure.

Findings

Binary linear classifiers are uniquely the most explainable classifiers.

Pointwise coverage effectively quantifies explainability.

The measure aligns with intuitive notions of simplicity and generalizability.

Abstract

We introduce the notion of pointwise coverage to measure the explainability properties of machine learning classifiers. An explanation for a prediction is a definably simple region of the feature space sharing the same label as the prediction, and the coverage of an explanation measures its size or generalizability. With this notion of explanation, we investigate whether or not there is a natural characterization of the most explainable classifier. According with our intuitions, we prove that the binary linear classifier is uniquely the most explainable classifier up to negligible sets.