Predicting numbers of errors using software science

Abstract

An earlier paper presented a model based on software science metrics to give quantitative estimate of the number of bugs in a programming project at the time validation of the project begins. In this paper, we report the results from an attempt to expand the model to estimate the total number of bugs to expect during the total project development. This new hypothesis has been tested using the data currently available in the literature along with data from student projects. The model fits the published data reasonably well, however, the results obtained using the student data are not conclusive.