I tried to submit a rating via CPANratings.perl.org, but I was unable to login (something is wrong with the server), so I'm submitting this here. This package is useful and appears to be amazingly bug-free for a 0.x version. There is one item that is missing that I would have expected to see. There should be a method that calculates the MSE (mean square error). I saw something by the author that suggests that he hasn't implemented this because he doesn't have a matrix inversion routine. Actually, the basic MSE formula does not require a matrix inversion. MSE= SSE / (n-p), where SSE is the sum of squared errors (sum of squared residuals), n is the number of data points, and p is the number of degrees of freedom in the model. I believe that the package must already be computing SSE, so the above calculation should be trivial. When the regression model is used for predicting new observations, then one may be interested in the estimated variance of the predictions; this formula does involve a matrix inversion, but the basic MSE formula would be an important addition. Dr. Phillip M. Feldman