Fuzzy regression: M-estimation approach

In this research, a new robust regression method is proposed which is based on employing a fitting criterion that is not as vulnerable as least-squares to unusual data. The proposed fuzzy regression in this paper is a modified Mestimation approach to fuzzy regression analysis, when independent variables are crisp, the dependent variable isfuzzy and outliers are present in the data set. The parameters estimation problem finally reduces to a weighted leastsquares approach to estimating the parameters of interest which is a simple approach both theoretically and computationally. To illustrate how the proposed method is applied numerical example is discussed and compared in methods from the literature.

Keywords:

Fuzzy regression , Weighted least-squares , Imprecise data