The effectiveness of three-way classification with interpretable perspective

As a typical methodology to deal with uncertain issues, the three-way decision (3WD) has been developed rapidly in nearly ten years, both in theories and applications. Three-way classification is one of the important research fields of 3WD, which utilizes the idea of 3WD to solve classification problems. In this paper, we focus on investigating the effectiveness of three-way classification through two evaluation indicators: the classification quality (Precision, Recall, Accuracy and F-1) and the decision cost. The comparisons between two-way classification and three-way classification are concretely analyzed, some mathematical properties, judging conditions and decision criteria of these two classification methods are also discussed in detail. Finally, the experimental results on eight UCI data sets validate the mathematical analysis, which reveal the effectiveness of three-way classification. (C) 2021 Elsevier Inc. All rights reserved.

Automated summary

Three-way decision (3WD) has rapidly developed as a typical methodology to deal with uncertain issues in the past decade. Three-way classification, as an important research field of 3WD, has been shown to effectively solve classification problems, with experimental results validating its effectiveness.