Bundling classifiers by bagging trees

The quest of selecting the best classifier for a discriminant analysis problem is often rather difficult. A combination of different types of classifiers promises to lead to improved predictive models compared to selecting one of the competitors. An additional learning sample, for example the out-of-bag sample, is used for the training of arbitrary classifiers. Classification trees are employed to bundle their predictions for the bootstrap sample. Consequently, a combined classifier is developed. Benchmark experiments show that the combined classifier is superior to any of the single classifiers in many applications. (c) 2004 Elsevier B.V. All rights reserved.