Error-type – A Novel Set of Software Metrics for Software Fault Prediction

In software development, identifying faults is an important task. The presence of not only reduces the quality software, but also increases cost development life cycle. Fault identification can be performed by analysing characteristics buggy source codes from past and predict present ones based on same using statistical or machine learning models. Many studies have been conducted to fault proneness systems. However, most them provide either inadequate insufficient information thus make prediction task difficult. this paper, we a novel set metrics called Error-type metrics, which provides models with about patterns different types Java runtime error. Particular, in study, ESM values consist three common errors are Index Out Of Bounds Exception, Null Pointer Class Cast Exception. Also, proposed methodology for modelling, extracting, evaluating error modules Stream X-Machine (a formal modelling method) techniques. experimental results showed that could significantly improve performances fault-proneness prediction.