Using Defect Tracking and Analysis to Improve Software Quality

Defect tracking is a critical component to a successful software quality effor t. In fact, Robert Grady of Hewlett-Packard stated in 1996 that “software defect data is [the ] most important available management information source for software process improveme nt decisions,” and that “ignoring defect data can lead to serious consequences for an organization’s business” [Grady96]. However, classifying defects can be a difficult t ask. As Ostrand and Weyuker paraphrased a 1978 report by Thibodeau, defect classification schemes of that time period often had serious problems, including “ambiguous, overlapping, and incomplete categories, too many categories, and confusion of error causes, fault symptoms, and actual faults.” [OstrandWeyuker84]. Yet the classific ation of defects is very important, and by examining the lessons learned by other organizations, one hopes to be in better position to implement or improve one’s own defect classification and analysis efforts. To this end, this report discusses five defect categorization and analysis efforts f rom four different organizations. This list of organizations should be taken as a sample of the range of schemes covered in research and in industry over the previous twenty-five years.