Boosting Oversampling Methods for Imbalanced Data Classification

Abstract

Imbalanced data classification is a prevalent challenge in supervised machine learning, where minority class instances are often underrepresented in the data set. This paper explores the efficacy of AdaBoost, a renowned ensemble learning method, in conjunction with two oversampling techniques, Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADASYN). In the proposed methodology, the data set is initially divided into training and testing sets. The training data is then divided into random subdata sets, each of which is used to develop unique models, which are commonly based on Decision Trees. These models are then weighted averaged together to generate the AdaBoost model. To address the issues of imbalanced data, misclassified instances from the testing set are extracted and combined to test set to produce a new data set. Both SMOTE and ADASYN oversampling techniques are applied to those misclassified data and their classification performance is compared using confusion matrix, accuracy, precision, and recall metrics. The purpose of this research is to provide insight into the suitability of AdaBoost in conjunction with oversampling strategies for dealing with imbalanced data classification. The experimental findings suggest that SMOTE is the best strategy for dealing with imbalanced data sets after oversampling rather than ADASYN.