Application of the theory of fuzzy sets for a qualitative analysis of mathematical models

Abstract

Abstract The article discusses the estimation of fuzzy regression parameters when specifying various membership functions, since the effectiveness of the least squares method dramatically reduces when a number of prerequisites for its use are violated, in particular, when the sample being processed contains observations that are poorly consistent with the others. In these cases, one can use estimation methods that are less sensitive than the least squares method to specification errors and allow one to obtain so-called fuzzy estimates. Among such methods is the method of smallest modules, the implementation of which leads to the problem of linear programming.