Can software project maturity be accurately predicted using internal source code metrics?

Abstract

Predicting a level of maturity (LoM) of a software project is important for multiple reasons including planning resource allocation, evaluating the cost, and suggesting delivery dates for software applications. It is not clear how well LoM can be actually predicted-mixed results are reported that are based on studying small numbers of subject software applications and internal software metrics. Thus, a fundamental problem and question of software engineering is if LoM can be accurately predicted using internal software metrics alone? We reformulated this problem as a supervised machine learning problem to verify if internal software metrics, collectively, are good predictors of software quality. To answer this question, we conducted a large-scale empirical study with 3, 392 open-source projects using six different classifiers. Further, our contribution is that it is the first use of feature selection algorithms to determine a subset of these metrics from the exponential number of their combinations that are likely to indicate the LoM for software projects. Our results demonstrate that the accuracy of LoM prediction using the metrics is below 61% with Cohen’s and Shah’s k<< 0.1 leading us to suggest that comprehensive sets of internal software metrics alone cannot be used to predict LoM in general. In addition, using a backward elimination algorithm for feature location, we compute top ten most representative software metrics for predicting LoM from a total of 90 software metrics.