Quantitative Quality Management through Defect

To produce high quality software, the final software should have as few defects as possible. The task of quality management in a software project is to plan suitable quality control activities, and properly execute and control these activities such that most of the defects are detected “in-process”, that is, before the software is delivered. In this article, we describe the approach of quantitative quality management through defect prediction and statistical process control that is employed at Infosys, a large ISO-certified software house that has been assessed to be at level 5 of the CMM. In this approach, a quality goal is set for a project in terms of the defect density delivered. To achieve the goal, the defect levels for different phases in the process are estimated using past data. During the execution of the project, the actual defect numbers are compared with the estimates to see if the project is progressing satisfactorily towards achieving the goal, or some correction is needed. To further improve the control and provide early warnings, the phase-wise control is complemented with activity-level control using statistical process control