Kombinasi Bootstrap Aggregation (Bagging) dan Information Gain untuk Mengatasi Ketidakseimbangan Kelas (Class Imbalance) pada Algoritma C4.5.

Abstract

Class imbalance is meant as one of the majority classes which is much more than the minority class. Consequently, this condition can cause the algorithm performance to decrease. This class imbalance can also affect the bad performance of C4.5 algorithm in classifying. Combination of feature selection method by using information gain and bagging can decrease class imbalance in the bad performance of C4.5 algorithm. The process of feature selection, using information gain is done to select attributes, using the threshold 0.02. The result of the accuracy of feature IG selection combination with bagging at the validation of 10 fold cross-validation can increase properly in classifying imbalanced data in diabetes data.