A study on the variables related to software quality: a data mining based approach

摘要

Advances in information technology have made possible accumulation of large amount of data arising from information systems. This fact enabled the emergence and growth of research on Data Mining (DM), which aims primarily at identifying new and useful patterns in databases. Recently, software quality has received attention among areas of application. The quality of software has become an important issue for the maintenance and customer attraction for businesses and can be considered a competitive weapon. On the other hand, the quality of decision making in project management software is closely related to information available about the process of organizational development. This information can be stored in data warehouses to be analyzed and interpreted using techniques of DM. This paper describes an application of DM bases of quality metrics for software products, which are aimed at identifying the variables of the development process that would mostly affect the error rate during the phases of testing, approval, and production. The study of the target company has a high level of standardization of processes. The results obtained in the models indicated a strong or moderate correlation of some variables of the process with product quality, which may be the target of management attention in order to: (ⅰ) improve the overall quality of the development process, (ⅱ) reduce the error rate in the systems developed and (ⅲ) promote greater customer satisfaction.