I have been using the NNET package and have successfully run neural networks on both continuous and binary targets. I managed to search the internet and found out how to capture the error resulting from a binary model no problems. My problem now is that I am trying to find how to calculate an approriate error when modelling a continuous target. As a statistician I immediately think of the RMSE, is it possible to calculate such a statistic from the model? This would require kniowledge of the degrees of freedom in the model (if there is an equivalent in a neural netwrok!?) Ideally I would like the proportion of the RMSE to total error. For example, one criteria of model fit may be no worse than 10% error rate. It is this sort of statistic that i am desperate to calc.Any help greatly appreciated.
[[alternative HTML version deleted]] ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.