OWA Based PCA Information Fusion Method for Classification Problemm

Abstract

Information plays an important role in modern enterprises, no matter in decision sup- porting and business strategies making all provide efficient supports to executive managers. However, information is getting more and more today, how to handle high dimensions and high complexities data are more difficult than before. This contribution presents a informa-tion fusion method based on the concepts of Ordered Weighted Averaging (OWA) operator and Principal Components Analysis (PCA) method to deal with above problems. The pro-posed method can be described briefly as follows: (1) Reduce data dimensions by PCA method. (2) Calculate integrated values by order OWA operator. (3) Cluster data instances into specific group and train classification accuracies to obtain the best situation parame-ter α. (4) Validate classification accuracies from testing data. This contribution employs five datasets to verify the performances of proposed method and the results of the experi-ments show that the proposed method actually surpasses the listing methods in classification accuracies.