Novel hybrid pair recommendations based on a large-scale comparative study of concept drift detection

作者：

Highlights：

• Pair-wise comparison of 15 detectors with 6 different classifiers accomplished.

• Proposes novel hybrid pairs to recommend the most efficient ones.

• Aims to explore whether well-known/cited ones yielded best results or novel ones.

• Using the Nemenyi post-hoc test, the best/worst pairwise configurations attained.

• As a comprehensive approach, towards detectors, paves the way for a meta-model.

摘要

•Pair-wise comparison of 15 detectors with 6 different classifiers accomplished.•Proposes novel hybrid pairs to recommend the most efficient ones.•Aims to explore whether well-known/cited ones yielded best results or novel ones.•Using the Nemenyi post-hoc test, the best/worst pairwise configurations attained.•As a comprehensive approach, towards detectors, paves the way for a meta-model.

论文关键词：Concept drift,Drift detection,Data stream,Classification,Pairwise comparison

论文评审过程：Received 8 July 2019, Revised 20 July 2020, Accepted 20 July 2020, Available online 25 July 2020, Version of Record 11 August 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113786