chenr86

I have tried p = 1.99 * ones(dim). Still, the model is overfitting and there is nearly no improvement. The predictive coefficient R squared in training set is 1 while...