Peningkatan Akurasi Prediksi Waktu Perbaikan Bug dengan Pendekatan Partisi Data

Abstract

Software developers need to have a plan in setting up software development costs. Software repairs in the system maintenance phase can be caused by bugs. Bugs are malfunctions that occur in software that does not meet the needs of the software. The software bug can have a fast or long time in the repair depending on the difficulty level. Developers can be assisted by predictive model recommendations and provide time-out considerations for bug fixes. Some research has been done on the predicted time of bug fixes using various existing classification algorithms with free datasets that can be accessed or downloaded from the software site. The classification of existing research uses several datasets of varying time ranges, the results obtained from a very variable time span are assessed to be further enhanced by partitioning over time ranges prior to classification. With partitions based on the repair timeframe, improved accuracy has been made with some classification methods. The results obtained after performing trials on multiple datasets are an increase for the majority of the datasets used. There is also a decrease in accuracy in some tests performed with a particular dataset.