From Uncertainty to Precision: Enhancing Binary Classifier Performance through Calibration

TL;DR

This paper emphasizes the importance of calibration in binary classifiers, introduces a new Local Calibration Score, and demonstrates how local regressions improve calibration and visualization, especially in credit default prediction.

Contribution

It introduces the Local Calibration Score and advocates for local regressions as effective recalibration and visualization tools in classifier performance assessment.

Findings

Local Calibration Score effectively measures calibration sensitivity.

Local regressions improve recalibration and visualization.

Application to credit default prediction demonstrates practical benefits.

Abstract

The assessment of binary classifier performance traditionally centers on discriminative ability using metrics, such as accuracy. However, these metrics often disregard the model's inherent uncertainty, especially when dealing with sensitive decision-making domains, such as finance or healthcare. Given that model-predicted scores are commonly seen as event probabilities, calibration is crucial for accurate interpretation. In our study, we analyze the sensitivity of various calibration measures to score distortions and introduce a refined metric, the Local Calibration Score. Comparing recalibration methods, we advocate for local regressions, emphasizing their dual role as effective recalibration tools and facilitators of smoother visualizations. We apply these findings in a real-world scenario using Random Forest classifier and regressor to predict credit default while simultaneously measuring calibration during performance optimization.