Predicting software build failure using source code metrics

In this paper, we describe the extraction of source code metrics from the Jazz repository and the application of data mining techniques to identify the most useful of those metrics for predicting the success or failure of an attempt to construct a working instance of the software product. We present results from a study using the J48 classification method used in conjunction with a number of attribute selection strategies applied to a set of source code metrics calculated from the code base at the beginning of a build cycle. The results indicate that only a relatively small number of the available software metrics that we considered have any significance for predicting the outcome of a build. These significant metrics are discussed and implication of the results discussed, particularly the relative difficulty of being able to predict failed build attempts. The results also indicate that there is some scope for predicting the outcomes of an attempt to construct a working instance of the software product by analysing the characteristics of the source code to be changed. This provides the opportunity for software project managers to estimate the risk exposure of the planned changes in the build prior to commencing the coding activities.