THE CHOQUET INTEGRAL WITH RESPECT TO λ-MEASURE BASED ON γ-SUPPORT
When the multicollinearity between independent variables occurs in the multiple regression models, its performance will always be poor. The traditional improved method which is always used is the ridge regression model. Recently, the Choquet integral regression model with fuzzy measure can further be exploited to improve this situation. In this study, we found that based on different fuzzy support, the Choquet integral regression model with the same fuzzy measure may have different performances, three kinds of fuzzy supports, C-support, V-support and γ-support proposed by our work were considered. For evaluating the performances of the Choquet integral regression models with P-measure or λ-measure based on above different fuzzy supports, a real data experiment by using a 5-fold cross-validation mean square error (MSE) is conducted. Experimental result shows that the Choquet integral regression model with λ-measure based on γ-support has the best performance.
Fuzzy measure fuzzy support C-support V-support γ - support
HSIANG-CHUAN LIU YU-CHIEH TU CHIN-CHUN CHEN WEI-SHENG WENG
Department of Bioinformatics, Asia University, Taiwan Graduate Institute of Educational Measurement and Statistics, National Taichung University, Taiwan Graduate Institute of Educational Measurement and Statistics, National Taichung University, Taiwan G
国际会议
2008 International Conference on Machine Learning and Cybernetics(2008机器学习与控制论国际会议)
昆明
英文
3602-3606
2008-07-12（万方平台首次上网日期，不代表论文的发表时间）