Penanganan Data Tidak Seimbang Menggunakan Hybrid Method Resampling Pada Algoritma Naive Bayes Untuk Software Defect Prediction

Abstract

Software defect prediction is software data that is used to identify a software module and can also be used to predict software defects. Before carrying out further trials, it is necessary to carry out special handling, especially by using algorithm models as predictions of software defects with the aim of obtaining information from the device being developed. Therefore, it is necessary to predict software defects using appropriate classification and prediction methods, so that the resulting accuracy results are better. In this study, the naïve Bayes algorithm was used as a classification with a resampling technique approach to handle unbalanced data, including SMOTEENN and SMOTETomek. The best accuracy results in the research conducted were 92.5% on the Nasa Repository PC4 dataset