طريقة ومعيار جديدان للكشف عن مشكلة التعدد الخطي ومعالجتها واختيار متغيرات الانحدار

Abstract

Multi co linearity is regarded as a big problem which may face a researcher while studying a linear relation between some random phenomenon and the factors that affect it using multiple linear regression equation because it leads to disfigure that relation for each factor such that the factors look like as if they where not important to explain the phenomenon. Several solutions are proposed to treat the problem but all these solutions- as some researchers believe- have side-effects such a disadvantaging at least one of (BLUE) properties and disabling to accomplish the adequate model as well as the simple explanation of the regression- coefficients. So the researcher proposed a new criterion and procedure to infer and treat the multicollinearity problem without side-effects approximately. He derived his criterion named RUF using OLS method to estimate the regression- parameters and he proved that criterion itself is the factor of changing the relative importance of regression variables more over he used it to achieve the best sub- regression model according to standard criteria and by comparing with the other procedures.

Key- words: