Is there a way to plot a confusion matrix of the cross validation results?

Can somebody tell me how to plot a confusion matrix of the crossval result?

 CVMdl = crossval(classifier,'HoldOut',0.08);
  k=kfoldLoss(CVMdl,'lossFun','classiferror','mode','average')
  L = resubLoss(classifier,'LossFun','classiferror')

   Accuracy = 1 - k

[ben dp]

Hey Hadeer!

Did you find a solution? I'm in the same problem.

[ROHAN JAIN]

Hi,

You can plot the confusion matrix easily by using the following function:

confusionchart(testlabels,labels_predicted)

where testlabels are the labels of the test set and labels_predicted refers to the labels that have been predicted by the LDA classifier using predict().

It automatically plots the confusion matrix. Further, you can also store it in a variable and access the values using the dot operator as mentioned below.

cvmat=confusionchart(testlabels,labels_predicted)
cval=cmat.NormalizedValues; % cval is the required matrix

Hope it helps!

Thanks

Answer 1

[Santhana Raj]

I am not aware of any method to plot confusion matrix. But usually I calculate the precision and recall from the true positives and true negatives. Some places I also use F-measure. Depending on your application, any of this might be a good measure to evaluate your classification algorithm.

Check wiki for the formulas for these.

Answer 2

[Karina Nanuck-Robertson]

Not sure if this helps

confusionmatrix