A calculation technology that eliminates the 'multicollinearity' of regression analysis and obtains a partial regression coefficient that indicates the degree of contribution of explanatory variables to appropriate objective variables and uses it as management data.

摘要

PROBLEM TO BE SOLVED: To provide a multiple regression analysis applied to an analysis of a relationship between an industrial result Y and a cause Xi, in many cases, the explanatory variable of the selected cause has a numerical value having a considerably high correlation with the objective variable Y. This makes the partial regression coefficient bi of the explanatory variable Xi, which indicates the degree of contribution to the result Y, different from theory and experience. SOLUTION: In order to solve this contradiction, a determinant of the correlation coefficient of the result Y of the normal equation of the multiple regression analysis and the explanatory variable Xi is created, and the numerical balance of both the left and right sides is not balanced. A new correlation determinant of the normal equation that removes the overlap with Xi from the dependent off-diagonal element Xj based on the diagonal element Xi, which is a variable, and uses the remainder as the off-diagonal element of the matrix. This is a calculation technique that eliminates "multicollinearity" and the like by finding the bi.