Using a new type of nonlinear integral for multi-regression: an application of evolutionary algorithms in data mining
暂无分享,去创建一个
We develop a nonlinear multi-regression model based on the Wang integral to describe a multi-input single-output system. In this model, in general, set function /spl mu/ is nonadditive. The nonadditivity of /spl mu/ describes the inherent interaction among the input attributes x/sub 1/, x/sub 2/, ..., x/sub n/. When the proper input-output data are available, by using the adaptive genetic algorithm shown in this paper, rather precise estimated values of parameter c, q, w and /spl mu/ of the regression model can be obtained. Thus, the multi-input single-output system can be used to make prediction. That is to say, when the values of input attributes x/sub 1/, X/sub 2/, ..., X/sub n/, are known, we can predict the output Y by calculating the nonlinear multi-regression.
[1] M. Sugeno,et al. Non-monotonic fuzzy measures and the Choquet integral , 1994 .
[2] G. Klir,et al. Fuzzy Measure Theory , 1993 .
[3] Kwong-Sak Leung,et al. A new type of nonlinear integrals and the computational algorithm , 2000, Fuzzy Sets Syst..