Threshold ML-KNN: Statistical Evaluation on Multiple Benchmarks

摘要

This paper concerns the performance of a recently proposed multi-label classification algorithm called Threshold ML-KNN. It is a modification of the established ML-KNN algorithm. The performance of both algorithms is compared on several publicly available benchmarks. Based on the results, the conclusion is drawn that Threshold ML-KNN is statistically significantly better in terms of accuracy, f-measure and hamming loss.