Fuzzy linear regression analysis

In this paper, a generalized fuzzy linear regression model Y=β1X1+β2X2+ëëë+βmXm where all βj and Xj (j=1,...,m) are fuzzy is considered. A new principle of estimating parameters, which is called Minmax principle, is introduced via the possibility distribution. The Minmax estimation of the regression model parameters and the least squares estimation of the approximation model parameters are discussed. Procedures of estimating these parameters are given.