Ensemble Methods: Boosting, Bagging and their Applications on Two Different Data Sets

Abstract

Bagging and boosting are ensemble learning methods that make groups more powerful when they come together and they are frequently used in data analysis. These methods combine a diversity of classifiers using the same learning algorithm for the base-classifiers. Ensemble methods also include logistic regression and linear discriminant analysis, effective on small and medium-sized data sets, and decision/regression trees, support vector machines, and artificial neural networks especially effective on large-sized data sets. However, the existence of a large number of classification methods brings with it the problem of selection. At this point, as an alternative, ensemble algorithms can be used to improve the performance of the selected classification method. The mostly referred ensemble methods are called “Bagging” and "Boosting”. This study aimed to show that how ensemble techniques work on different data sets and the results about bagging and boosting algorithms were evaluated on two different data sets.