Abstract
In the possiblistic regression analysis, the obtained fuzzy mean values bu applying to various data seem alike as to the coefficients obtained by normal multi regression analysis.On the other hand, the spreads of fuzzy coefficients are generally determined by including entire independent variables subject to minimize the summation of spreads.However, sometimes it happens to that it is very hard to explain the spreads rationally.Why does this happen and what is needed to improve this are investigated through simple examples.And real data obtained from experimental testes are also analyzed by proposed method in this paper.
