next up previous
Next: AdaBoost Applet [3 P] Up: MLA_Exercises_160106 Previous: Weak-Learner with Weighting [3

AdaBoost [3 P]

Analyze the properties of the AdaBoost algorithm with the WEKA toolkit. Use credit-g.arff from the UCI collection of datasets to test the classifiers.
Split your dataset into a training set containing 75% of the instances, and a test set with 25% of the instances.
Train a DecisionStump classifier on the dataset.
Use AdaBoostM1 to improve the results of the decision stump. Evaluate how the number of iterations influences the training and test error.
Plot the training and test errors of AdaBoost for at least the first 10 iterations. Interpret the results.

Pfeiffer Michael 2006-01-18