TCE: A Test-Based Approach to Measuring Calibration Error
This work addresses the need for reliable calibration assessment in machine learning, particularly for imbalanced data, but it is incremental as it builds on existing calibration error metrics.
The paper tackles the problem of measuring calibration error in probabilistic binary classifiers by proposing a new metric called test-based calibration error (TCE), which offers clear interpretation, consistent scale unaffected by class imbalance, and enhanced visual representation, as demonstrated through experiments on real-world imbalanced datasets and ImageNet 1000.
This paper proposes a new metric to measure the calibration error of probabilistic binary classifiers, called test-based calibration error (TCE). TCE incorporates a novel loss function based on a statistical test to examine the extent to which model predictions differ from probabilities estimated from data. It offers (i) a clear interpretation, (ii) a consistent scale that is unaffected by class imbalance, and (iii) an enhanced visual representation with repect to the standard reliability diagram. In addition, we introduce an optimality criterion for the binning procedure of calibration error metrics based on a minimal estimation error of the empirical probabilities. We provide a novel computational algorithm for optimal bins under bin-size constraints. We demonstrate properties of TCE through a range of experiments, including multiple real-world imbalanced datasets and ImageNet 1000.