Study On Multicollinearity In Linear Regression Model

In the practical application of multivariate linear regression model,there is almost no completely unrelated phenomenon existed in explanatory variables.Especially in the study of certain economic problems,which often involves multiple variables,the widespread phenomenon of multicollinearity existed in each variable,which will cause certain influence to the model.It will not only affect the estimation of the parameters,but also expand the error of the modle,resulting in destruction of the model's stability.Therefore,this paper will discuss the above mentioned problem.Now,with the scientific progress and development,people have explored many methods of theory and different functions to solve the multicollinearity problem.Their advantages and disadvantages provide a strong basis for people to choose the appropriate specific model.This paper mainly introduces two kinds of methods: ridge regression method and principal component regression method.The combination of theory and practice explained the interpretation of the two methods.First of all,the paper starts from the theoretical analysis of the basic concept of the two methods;secondly,the paper illustrates the different roles and characteristics of the two models through specific examples.In addition,this paper also introduces an improved method of generalized ridge estimation and ridge regression through the analysis.According to the calculation of the mean square error and mean square error of the numerical size of the model,people can choose a better model to solve the problem.