Variable selection in the linear regression model

Finding the appropriate subset of independent variables for the model in multiple linear regression is called the variable selection problem. It is an important problem to decide which variables will be included in the model and which variables will be excluded from the model, in order that the effects of the independent variables on the dependent variable are significant and can explain the dependent variable. For this purpose, some criteria used in the selection of the best model in the multiple linear regression model were firstly examined. Then, in the multiple linear regression model, in case of multicollinearity and/or outliers in the data, some criteria used in model selection were discussed. Key Words: Model selection, Multicollinearity, Outlier, Biased estimator, Robust Estimator