Contents
Description of demo_multiclass_boosting.m
Demonstrates boosted stump regression using the AdaBoost algorithm for a multiclass classification problem
close all clear all generateData_4grid
usage of bagged decision trees
options_bag = []; options_bag.subModel = @ml_multiclass_stump; model_bag = ml_multiclass_bagging(Xtrain, ytrain, options_bag); yhat_bag = model_bag.predict(model_bag, Xtest); testError_bag = mean(yhat_bag ~= ytest); fprintf('Averaged misclassification test error with %s is: %.3f\n', ... model_bag.name, testError_bag);
Averaged misclassification test error with Classification with Bagging is: 0.280
usage of boosted multi-class stump regression with AdaBoost
options_bs1 = []; options_bs1.nBoosts = 1000; options_bs1.booster = 'ada'; options_bs1.subModel = @ml_multiclass_stump; model_bs1 = ml_multiclass_boosting(Xtrain, ytrain, options_bs1); yhat_bs1 = model_bs1.predict(model_bs1, Xtest); testError_bs1 = mean(yhat_bs1 ~= ytest); fprintf('Averaged misclassification test error with %s is: %.3f\n', ... model_bs1.name, testError_bs1)
Averaged misclassification test error with AdaBoosted Classification is: 0.164
figure; plotClassifier(Xtrain, ytrain, model_bag); figure; plotClassifier(Xtrain, ytrain, model_bs1);

