Software quality control and prediction model for maintenance

We develop a quality control and prediction model for improving the quality of software delivered by development to maintenance. This model identifies modules that require priority attention during development and maintenance by using Boolean discriminant functions. The model also predicts during de velopment the quality that will be delivered to maintenance by using both p oint and confidence interval estimates of quality. We show that it is impor tant to perform a marginal analysis when making a decision about how many m etrics to include in a discriminant function. If many metrics are added at once, the contribution of individual metrics is obscured. Also, the margina l analysis provides an effective rule for deciding when to stop adding metr ics. We also show that certain metrics are dominant in their effects on cla ssifying quality and that additional metrics are not needed to increase the accuracy of classification. Related to this property of dominance is the p roperty of concordance, which is the degree to which a set of metrics produ ces the same result in classifying software quality. A high value of concor dance implies that additional metrics will not make a significant contribut ion to accurately classifying quality; hence, these metrics are redundant. Data from the Space Shuttle flight software are used to illustrate the mode l process.