Model selection in multiple regression by applying genetic algorithm and by using information criteria

The number of models increases exponentially when the explanatory variables increases in a multiple linear regression model. In this case, model selection is impossible by using traditional procedures, stepwise methods and even existing statistical softwares. In this study, the model selection problem in a multiple linear regression model when there are more explanatory variables or regressors is considered by applying genetic algorithm and by using information criterias. For this purpose first, general information about multiple linear regression model are given and building multiple linear regression model is explained. Then, the best model selection problem in a multiple linear regression model when there are more explanatory variables is examined by stepwise methods. After than, genetic algorithm and information criterias for multiple linear regression model are emphasized, following model selection in multiple regression by applying genetic algorithm and by using information criterias is explained. Finally, results and discussions are given. Key words: Information criterion, Multiple linear regression, Genetic algorithm, Model selection.