Abstract
In this paper, interval regression methods are proposed for a realistic data set which has some outliers.The proposed methods are generalization of the interval regression method proposed by Tanaka et al., and have the possibility of wide application to the fuzzy or possibilistic regression analysis.First a mixed 0-1 integer programming problem is formulated to obtain the interval regression model which includes k data points out of m given data points.A computational algorithm of this problem is proposed.Next a multi-objective programming problem is formulated to cope with the situation where the number of data points included in the interval regression model is not decided in advance.This problem has two objectives : to minimize the width of the model and to maximize the number of data points included in it.The effectiveness of the proposed methods is demonstrated by numerical examples.
