Modeling factors for improving correctness of software systems: an empirical approach

One of the important aspects of software that determines its quality is correctness. A buggy system, even though might be functionally correct to meet the specifications, could essentially result in excessive maintenance costs. Although correctness of a system depends on several factors, the number of errors in the system plays a crucial role in affecting its correctness. Therefore, this paper focuses on the number of errors present in software, as an indicator of correctness. This paper presents the results of a study aimed at investigating the relationships between different software measures and the number of errors present. The results of the study should help one in controlling the number of errors from the early phases of development. It was discovered that most of the metrics considered in the study among many other metrics, have strong relationship with errors. On the other hand, few metrics did not demonstrate remarkable relationship.