Abstract: Multicollinearity problem in learning machines occurs when there are high dependencies among the input variables. The problem increases the variance of predictive model to cause unstable results. In regression models, the multicollinearity is also a problem to be solved. Ridge regression is a good method to settle the problem of regression. In general, the shrinkage parameter of ridge regression is determined by the arts of researchers. But, the selections are not always good. So, in this study, we propose an improvement of ridge regression using differential evolution. This is an evolutionary ridge regression to find better shrinkage parameter. To verify performance of our research, we make experiments using objective data sets.