Improvisasi Teknik Oversampling MWMOTE Untuk Penanganan Data Tidak Seimbang
Imbalance data is a condition which there is a distinction in the quantity of data that results withinside the majority class (classes with very many members) and minority class (classes with very few members). It can complicate the classification process since the machine learning algorithm method is designed to classify already balanced data. The oversampling process technique is used to resolve data imbalance by applying synthetic data to the minority class in such a manner that it has the same volume of data as the majority class. MWMOTE is an oversampling technique that generates synthetic data based on members of the minority class clusters that are close to the majority class. This approach is capable of generating synthetic data well. The resulting synthesis data remains in the nearby majority region and too dense on the border of the cluster. It is hence permitting the resulting synthetic data to go into the majority class classification. This study is objectives to improve the process of generating synthetic data on MWMOTE so that the resulting data is extensively dispensed withinside the minority class. The outcomes of the test show that the proposed method is capable of enhancing the classification performance for KNN and C4.5 Decision Tree classification sequentially by 0.46% and 0.96% compared to MWMOTE