What is the cut-off point of MSE in neuralnet in R. Most of the resources I came across stated that a value closer to zero is perfect. So, what is the upper limit? Are values >= 10, 20 or 30 acceptable?
I do not think that there is any hard and fast cut-off point in mean square error (MSE). It depends on the researcher, the sample used and the circumstances. The MSE is a function of the sample size; the bigger the sample, the less it is and the less it is, the more precise the estimates.That is why it is always advised that we draw a sample as large as possible. MSE of value 10, 20 or 30 could be acceptable in some contexts and unacceptable in others.