抄録
A fuzzy regression model illustrates the possibilities of the target system by including all data in the model. Therefore, samples influence and distort the shape of the interval model such as a fuzzy regression model if they are far from the center of data. Yabuuchi and Watada have proposed two fuzzy robust regression models. The first model employs distance between the boundary of model and irregular samples in order to remove an influence of irregular samples. The second model is built by maximizing sum of a possibility grade between the model and data in order to coincide with data distribution. The second fuzzy robust regression model is focused on in this paper. Although this fuzzy robust regression model has been improved a conventional fuzzy regression model, there are some problems to solve. The objective of this paper is to confirm these problems.
