Predicting risk of software changes

Reducing the number of software failures is one of the most challenging problems of software production. Weassume that software development proceeds as a series of changes and model the probability that a change tosoftware will cause a failure. We use predictors based on the properties of a change itself. Such predictorsinclude size in lines of code added, deleted, and unmodified; diffusion of the change and its componentsubchanges, as reflected in the number of files, modules, and subsystems touched, or changed; severalmeasures of developer experience; and the type of change and its subchanges (fault fixes or newcode). The model is built on historic information and is used to predict the risk of new changes. In thispaper we apply the model to 5ESS ® software updates and find that change diffusion and developerexperience are essential to predicting failures. The predictive model is implemented as a Web‐basedtool to allow timely prediction of change quality. The ability to predict the quality of change enables us tomake appropriate decisions regarding inspection, testing, and delivery. Historic information on softwarechanges is recorded in many commercial software projects, suggesting that our results can be easily and widelyapplied in practice.