How can I get the R-values after training a neural network ?

I created a neural network but to compare different networks, I want to have the R-value . When I trained the network with the nntraintool I can click on regression and I get a nice graph with the regression plots including the R-values.
Now I want to put those values in a matrix. I have already tried the following code: [m,b,r]=postreg(targets,outputs) to get the overall R-value. But my error is: "A and T cell arrays must have 1, 2, or 3 elements."
When I try to create a regression plot on my own with the code: plotregression(targets,outputs). I receive a good plot, equal to the one of the nntraintool but the R, m and b-values are NaN.
What is the problem or can I get the R-value on another way?
Thanks in advance!

 Respuesta aceptada

Greg Heath
Greg Heath el 19 de Sept. de 2012
help regression
doc regression
Hope this helps.
Thank you for officially accepting my answer.
Greg

Más respuestas (1)

ThijsV
ThijsV el 19 de Sept. de 2012
Thanks,
I think my problem is that my row vector of the targets contains NaN values but the output of my network not.
Is there an easy way so my NaN values are not taken into account?
Kind regards,
Thijs

Categorías

Más información sobre Deep Learning Toolbox en Centro de ayuda y File Exchange.

Preguntada:

el 19 de Sept. de 2012

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by