Identification of possibilistic coefficients in fuzzy linear systems

This paper proposes possibilistic regression analysis by the identification method of possibility distribution. First we obtain all possible coefficient vectors of linear systems which correspond to different sets of data. After selection of the necessary coefficient vectors, we apply the identification method to the selected coefficient vectors to obtain a possibility distribution of coefficients in a fuzzy linear system. By our proposed approach, the smaller area of possibility distribution can be obtained. This is the characteristic of our proposed regression, compared with the other regression analyses