English

Block-regularized 5$\times$2 Cross-validated McNemar's Test for Comparing Two Classification Algorithms

Machine Learning 2025-01-07 v2

Abstract

In the task of comparing two classification algorithms, the widely-used McNemar's test aims to infer the presence of a significant difference between the error rates of the two classification algorithms. However, the power of the conventional McNemar's test is usually unpromising because the hold-out (HO) method in the test merely uses a single train-validation split that usually produces a highly varied estimation of the error rates. In contrast, a cross-validation (CV) method repeats the HO method in multiple times and produces a stable estimation. Therefore, a CV method has a great advantage to improve the power of McNemar's test. Among all types of CV methods, a block-regularized 5×\times2 CV (BCV) has been shown in many previous studies to be superior to the other CV methods in the comparison task of algorithms because the 5×\times2 BCV can produce a high-quality estimator of the error rate by regularizing the numbers of overlapping records between all training sets. In this study, we compress the 10 correlated contingency tables in the 5×\times2 BCV to form an effective contingency table. Then, we define a 5×\times2 BCV McNemar's test on the basis of the effective contingency table. We demonstrate the reasonable type I error and the promising power of the proposed 5×\times2 BCV McNemar's test on multiple simulated and real-world data sets.

Cite

@article{arxiv.2304.03990,
  title  = {Block-regularized 5$\times$2 Cross-validated McNemar's Test for Comparing Two Classification Algorithms},
  author = {Jing Yang and Ruibo Wang and Yijun Song and Jihong Li},
  journal= {arXiv preprint arXiv:2304.03990},
  year   = {2025}
}

Comments

12 pages, 6 figures, and 5 tables

R2 v1 2026-06-28T09:55:24.941Z