Binary Classifier Calibration using an Ensemble of Near Isotonic Regression Models

TL;DR

This paper introduces ENIR, a novel ensemble calibration method for binary classifiers that improves probability estimates by addressing isotonic regression limitations, demonstrating superior performance on synthetic and real datasets.

Contribution

ENIR extends existing calibration methods by relaxing the monotonicity constraint, providing a more flexible and accurate calibration approach for binary classifiers.

Findings

ENIR outperforms several calibration methods on synthetic datasets.

ENIR significantly improves calibration on real-world datasets.

The method is computationally efficient at O(N log N) time complexity.

Abstract

Learning accurate probabilistic models from data is crucial in many practical tasks in data mining. In this paper we present a new non-parametric calibration method called \textit{ensemble of near isotonic regression} (ENIR). The method can be considered as an extension of BBQ, a recently proposed calibration method, as well as the commonly used calibration method based on isotonic regression. ENIR is designed to address the key limitation of isotonic regression which is the monotonicity assumption of the predictions. Similar to BBQ, the method post-processes the output of a binary classifier to obtain calibrated probabilities. Thus it can be combined with many existing classification models. We demonstrate the performance of ENIR on synthetic and real datasets for the commonly used binary classification models. Experimental results show that the method outperforms several common binary classifier calibration methods. In particular on the real data, ENIR commonly performs statistically significantly better than the other methods, and never worse. It is able to improve the calibration power of classifiers, while retaining their discrimination power. The method is also computationally tractable for large scale datasets, as it is $O(N \log N)$ time, where $N$ is the number of samples.