You just square r. The Pearson's correlation tells you if the regression model is a good one. Since then . If the Pearson's correlation is close to 1, then that mean's that the model is a good model. If the person's correlation is not close to 1, then it means it is not a good model and you should use a different regresssion model.