Difference between regression and performance plot of Artificial neural network in MATLAB

Question

I am having problem understanding regression and performance plots of ANN. My data consists of 13 inputs and 3 outputs. Parameters used for simulation are as follows.

The problem I am facing is that I get a very good fitted regression plot as follows (Performance and regression plots).

This means that the model is well fitted, but on the other hand the mean squared error is quite high. Which is strange thing for me. If the regression plot is well fitted then the MSE should be small, or I am wrong?

Answer 1

The mean squared error has to be seen in relation to the size of the values you are fitting. It seems your values are on the order of 10^7, and your MSE is on the order of 10^10. This means, after taking the square root, your errors have size on the order of 10^5, which is about 1% error when viewed in relation to the magnitude of your target values.