The Effect of Class Imbalance Against LVQ Classification

Abstract

Accuracy is a measure of the capability of an algorithm, and studies use different classification methods to improve this benchmark. However, improper data collection adversely affects accuracy. In this study, we discuss how to influence the accuracy of data collection mechanisms. The learning vector quantization (LVQ) algorithm is tested to determine the effect of data sampling on accuracy. Training and test data are gathered in the data collection process. Results show that sampling techniques and retrieval of training and test data influence the accuracy of the LVQ classification method. Therefore, the chosen sampling technique can improve accuracy relative to overall data usage.