Hybrid Approach Redefinition (HAR) Method with Loss Factors in Handling Class Imbalance Problem

Class imbalance is the main problem in classification because the classification process tends to misclassify minority class which is an interesting class in another class if the training process is done to a set of instances. This problem will result in the result obtained biased towards to the class with a large number of instances. Against a number of methods proposed to overcome this class imbalance problem. One good method is the Hybrid Approach Redefinition (HAR) Method which has the advantage in overcoming the problem of class imbalance with the number of small classifiers and also the data diversity is good. This study will use the HAR Method incorporated with Loss Factors to correct the classification of most classes based on the performance evaluation of each classifier based on the F-Measure and G-Mean values. The results showed that HAR Method with Loss Factors gave better performance value compared with HAR Method without Loss Factor.