A New Fuzzy Supervised Classification Method Based on Aggregation Operator

Abstract

A new fuzzy supervised classification method based on aggregation operator is proposed in the present article. The proposed classifier aggregates the information extracted by exploring feature-wise degree of belonging to classes. I uses a pi-type membership function and MEAN (average) aggregation reasoning rule (operator). The effectiveness of the proposed classifier is verified with four benchmark data sets including a realtime financial domain data. Various performance measures are used for quantitative evaluation of the classifier. Experimental results on these data sets illustrate significant improvement in the classification performance of the proposed method compared to three other fuzzy classifiers, namely, explicit fuzzy, fuzzy k-nearest neighbor and fuzzy maximum likelihood.