An empirical study of a three-group software quality classification model

Timely and accurately predicting quality of software systems is an important issue in software reliability engineering. Software quality estimation models based on software metrics provide a systematic and scientific way to detect fault-prone modules. With the aid of such models, software developers can achieve high quality in software systems by focusing their endeavors on most faulty modules within limited resources and budget. Previous works related to classification models for software quality usually classified modules into two groups, fault-prone and not fault-prone. This paper presents a new technique for classifying modules into three groups, i.e., high-risk, medium-risk, and low-risk based on their rank-order associated with a quantitative quality factor, such as number of faults. This new technique calibrates three-group models with consideration to the resources available, which makes it different from other classification techniques. The proposed three-group classification method proved to be efficient and useful for resource utilization in software quality control.