I have now noticed in two independent projects with two independent developers that the RSME values in nnet for regression tasks (not classification) are significantly higher than, for example, a linear regression or the direct implementation in the nnet package. Something seems to be going wrong, but I have not yet found the cause.
I have now noticed in two independent projects with two independent developers that the RSME values in nnet for regression tasks (not classification) are significantly higher than, for example, a linear regression or the direct implementation in the nnet package. Something seems to be going wrong, but I have not yet found the cause.