MODIFIED METHOD OF CONSTRUCTING A MULTIVARIATE LINEAR REGRESSION GIVEN BY A REDUNDANT DESCRIPTION

A number of scientific works Prof. O. A. Pavlov and his disciples is devoted to the development an original method efficient estimation coefficients at nonlinear terms multivariate polynomial regression given by a redundant description under conditions active experiment. The solution formulated problem reduced sequential construction univariate regressions (finding estimates for terms) solving corresponding systems linear nondegenerate equations, variables which are description. Thus, was in We have proposed its that uses cluster analysis algorithm. algorithm’s implementation significantly reduces enumeration partial descriptions followed finding residual sum squares each them. This allows using chi-squared criterion build linguistic variable value gives qualitative assessment (high reliability, acceptable low unreliability) obtained result. computational experiments made it possible modify method, increased efficiency, first all, reliable structure sought modification, particular, has led more use general procedure least method.