Özet:

Abstract Software engineering research is centered on defect prediction. Successful software development requires better communication between data mining and software engineering. Software defect prediction is a pre-testing technique that estimates where bugs will show up in the code. The purpose of software defect prediction research is to identify potentially flawed parts of a programme before it reaches the testing phase. The primary benefit of these prediction models is that they need more testing time and money. may be directed to the modules most prone to errors. However, only a few mobile app-specific software defect prediction algorithms currently exist. It is common practise to utilise defect prediction algorithms to probe the impact domain in software (clustering, neural networks, statistical methods, and machine learning models). This research aims to examine and compared various ML (machine learning) algorithms for software bug prediction. Despite the widespread availability of failure prediction methods, no one strategy is appropriate for every data collection. Support Vector Machine, Random Forest, Naive Bayes, Logistic Regression, and Artificial Neural Network, were only some of ML methods utilised to find biggest possible subset of faults. The goal of this study is to utilize 5 data sets (JM1, KC1, KC2, PC1, and DS1) to identify flaws. As compared to other methods, ANN has been demonstrated to have the highest accuracy (93.8%).

Anahtar Kelimeler: