A Python Package itca for Information-Theoretic Classification Accuracy: A Criterion That Guides Data-Driven Combination of Ambiguous Outcome Labels in Multiclass Classification

Abstract

The itca Python package offers an information-theoretic criterion to assist practitioners in combining ambiguous outcome labels by balancing the tradeoff between prediction accuracy and classification resolution. This article provides instructions for installing the itca Python package, demonstrates how to evaluate the criterion, and showcases its application in real-world scenarios for guiding the combination of ambiguous outcome labels.