This page holds the press statistic formula for calculating prediction residual error sum of squares in r. As per the predictive error sum of squares formula given below, PRESS statistic has a common disadvantage, which they are dependent on the data size, and that you cannot compare single model's performance against data sets of multiple models. The lower value of predictive error sum of square is always desirable for a better predictive model.

The PRESS statistic in r calculation is always calculated based on the leave-one-out (l-o-o) for linear models, which is also termed as the cross-validation. The predictive residual error sum of squares statistics calculations are dependent on data sizes.