Year: 2025
Type: preprint
Abstract: Two novel, simple metrics for classifier performance are proposed: An accuracy-based R 2 and its scaled variant, the fair accuracy. They both measure the performance relative to the uninformed classifier and are better suited for imbalanced classes than common metrics, like the F 1 score or MCC.
Author: Igor Fischer
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4413638597