Fuzzy regression: a genetic programming approach

Given some data pairs (X~/sub i/, Y~/sub i/), 1/spl les/i/spl les/k, of fuzzy numbers, we are interested in finding a fuzzy function F which best fits the given data. Because of fuzzy arithmetic, we cannot compute a fuzzy function with F(X~/sub i/)=Y~/sub i/ for all i, as in the crisp case. Therefore, we used a genetic programming approach to find a suitable fuzzy function. We present some tests and argue that this method is quite suitable for obtaining a fuzzy function which can explain the given data.

[1]  Hideo Tanaka,et al.  Fuzzy linear regression combining central tendency and possibilistic properties , 1997, Proceedings of 6th International Fuzzy Systems Conference.

[2]  Lei Zhang,et al.  A Fuzzy Regression Method Based on Genetic Algorithm , 1994 .

[3]  Phil Diamond,et al.  Fuzzy least squares , 1988, Inf. Sci..

[4]  Abraham Kandel,et al.  General fuzzy least squares , 1997, Fuzzy Sets Syst..

[5]  James J. Buckley,et al.  Multivariate Non-Linear Fuzzy Regression: An Evolutionary Algorithm Approach , 1999, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[6]  James J. Buckley,et al.  Linear and non-linear fuzzy regression: Evolutionary algorithm solutions , 2000, Fuzzy Sets Syst..

[7]  M. Inuiguchi,et al.  Mean-absolute-deviation-based fuzzy linear regression analysis by level sets automatic deduction from data , 1997, Proceedings of 6th International Fuzzy Systems Conference.

[8]  H. Aso,et al.  Fuzzy regression analysis using RFLN and its application , 1997, Proceedings of 6th International Fuzzy Systems Conference.