Train a classification tree classifier, and then cross-validate it using a customk-fold loss function.

Load Fisher’s iris data set.

loadfisheriris

Train a classification tree classifier.

Mdl = fitctree(meas,species);

Mdlis aClassificationTreemodel.

Cross-validateMdlusing the default 10-fold cross-validation. Compute the classification error (proportion of misclassified observations) for the validation-fold observations.

rng(1);% For reproducibilityCVMdl = crossval(Mdl); L = kfoldLoss(CVMdl,'LossFun','classiferror')

L = 0.0467

Examine the result when the cost of misclassifying a flower asversicoloris10, and the cost of any other misclassification is1. Create the custom functionnoversicolor(shown at the end of this example). This function attributes a cost of10for misclassifying a flower asversicolor, and a cost of1for any other misclassification.

Compute the mean misclassification error with thenoversicolorcost.

mean(kfoldfun(CVMdl,@noversicolor))

ans = 0.2267

This code creates the functionnoversicolor.

functionaverageCost =也不要sicolor(CMP,~,~,~,Xtest,Ytest,~)% noversicolor Example custom cross-validation function% Attributes a cost of 10 for misclassifying versicolor irises, and 1 for% the other irises. This example function requires the fisheriris data% set.Ypredict = predict(CMP,Xtest); misclassified = not(strcmp(Ypredict,Ytest));% Different resultclassifiedAsVersicolor = strcmp(Ypredict,'versicolor');% Index of bad decisionscost = sum(misclassified) +...9*sum(misclassified & classifiedAsVersicolor);% Total differencesaverageCost = cost/numel(Ytest);% Average errorend