A comparison on Choquet integral with respect to different information-based fuzzy measures

In this paper, for grouped data, three kinds of the Choquet integral regression models with fuzzy measures based on joint entropy, complexity and multiple mutual information is considered. The above three fuzzy measures are called, E-measure, C-measure and M-measure, respectively. For evaluating the Choquet integral regression models with these three information-based fuzzy measures, a real grouped data experiment by using a 5-fold cross validation accuracy is conducted. The performances of the Choquet integral regression models based on these three fuzzy measures, respectively, and the traditional multiple linear regression model are compared. Experimental result shows that the Choquet integral regression model based on our proposed M-measure has the best performance and it outperforms the Choquet integral regression model based on our previous proposed C-measure.