Analisis Hasil Klasifikasi CBA Untuk Data Imbalallce

Abstract

Dataset is imbalanced if the classes are not approximately equally represented. There are many ways to make a dataset in a balance condition. SMOTE or Synthetic Minority Oversarnpling Technique is one of the oversampling techniques for increasing performance in imbalance dataset using synthetic samples. The accuration of the balance dataset produced by SMOTE can be measured using Classification Based on Association (CBA). CBA is a classification technique that combines classification rule mining method and association rule mining method. The combining of classification and association method causing the result of CBA represented in rule form. The Chalwa research in 2002 shows that SMOTE can increase the accuration of classification result in classifier that build from non-association based classification technique. This research goal is trying to find whether an association based classifier like CBA can improve the accuration in dataset that produced by SMOTE. The comparison is done by using imbalance dataset without using SMOTE process and dataset with SMOTE process, to see the effect of SMOTE in the accuration level and the number of rules produced.