Model selection using the genetic algorithm
AIC; BIC; EDC; model selection
Many practical problems involving linear models has a step that consists in reducing the number of variables of the model, either it is very expensive to deal with too many variables or because some of the variables are able to explain the response satisfactorily. We can cite among such methods of reducing the number of variables of a linear model, the principal component analysis, best subset selection, forward stepwise selection, etc. In this work, we present how to use the elitist genetic algorithm in order to select a collection of variables for a linear model.