MATLAB Answers

0

how to get confusion matrix when neural network toolbox is used?

Asked by pranjal on 20 Jun 2015
Latest activity Commented on by B.k Sumedha on 25 Jun 2015
i have used NN toolbox for classification. after network is trained, "performance plot", "regression plot" and "training state plot" options are displayed. there is no option for displaying confusion matrix. how can i get it?

  0 Comments

Sign in to comment.

2 Answers

Answer by Greg Heath
on 21 Jun 2015
 Accepted Answer

Feedforwardnet is a generic net that NEVER has to be called directly.
Use fitnet for regression/curvefitting and use patternnet for classification/pattern-recognition. Both call feedforwardnet internally with the appropriate options.
Using patternnet, the plot options are
Performance
Training State
Error Histogram
Confusion
Receiving Operating Characteristic
Hope this helps.
Thank you for formally accepting my answer
Greg

  0 Comments

Sign in to comment.


Answer by B.k Sumedha on 20 Jun 2015

As soon as you train the network you get the option to plot the confusion matrix.Take a look at this pdf.

  4 Comments

Show 1 older comment
I think you need to use certain commands like given in below document. Upto my knowledge you cant directly plot the matrix just my using nntool. http://in.mathworks.com/help/nnet/ref/plotconfusion.html
Of course you can
[x,t] = simpleclass_dataset;
net = patternnet;
net = train(net,x,t);
y = net(x);
The nntool automatically pops up with the following option buttons
Performance
Training State
Error Histogram
Confusion
Receiver Operating Characteristic
Just click on the button
Hope this helps.
Thank you for formally accepting my ABOVE answer
Greg
Yes and its obvious that it must pop-up. But just using the
'nntoool' command window you cant get those buttons which you have specified.

Sign in to comment.