An Ensemble Learning Approach Based on Missing-Valued Tables
Seiki Ubukata, Taro Miyazaki, Akira Notsu, Katsuhiro Honda, Masahiro Inuiguchi · Lecture notes in computer science · 2015
In classification problems on rough sets, the effectiveness of ensemble learning approaches such as bagging, random forests, and attribute sampling ensemble has been reported. We focus on occurrences of deficiencies in columns on the original decision table in random forests and attribute sampling ensemble approaches. In this paper, we generalize such deficiencies of columns to deficiencies of cells and propose an ensemble learning approach based on missing-valued decision tables. We confirmed the effectiveness of the proposed method for the classification performance through numerical experiments and the two-tailed Wilcoxon signed-rank test. Furthermore, we consider the robustness of the method in absences of condition attribute values of unknown objects.