An empirical study on predicting defect numbers

Abstract

Abstract-Defect prediction is an important activity to make software testing processes more targeted and efficient. Many methods have been proposed to predict the defect-proneness of software components using supervised classification techniques in within-and cross-project scenarios. However, very few prior studies address the above issue from the perspective of predictive analytics. How to make an appropriate decision among different prediction approaches in a given scenario remains unclear. In this paper, we empirically investigate the feasibility of defect numbers prediction with typical regression models in different scenarios. The experiments on six open-source software projects in PROMISE repository show that the prediction model built with Decision Tree Regression seems to be the best estimator in both of the scenarios, and that for all the prediction models, the results yielded in the cross-project scenario can be comparable to (or sometimes better than) those in the within-project scenario when choosing suitable training data. Therefore, the findings provide a useful insight into defect numbers prediction for those new and inactive projects.

FSCR:A Feature Selection Method for Software Defect Prediction

“…Graves et al [18] presented a generalized linear regression based method for the number of faults prediction using various change metrics datasets collected from a large telecommunication system and found that modules age, changes made to module and the age of the changes were significantly correlated with the defect-prone. Chen et al [11] performed an empirical study on predicting the number of faults using six regression algorithms and found that the prediction model built with decision tree regression had the highest prediction accuracy in most cases. In another similar study, Rathore et al [9] presented an experimental study to evaluate and compare the other six regression algorithms for the number of faults prediction.…”

“…The reason we choose these regression models is that these models perform best in predicting the number of software faults [11][12].…”

“…Some researchers [7][8][9][10][11][12] have investigated genetic programming, decision tree regression, and multilayer perceptron in the context of the number of faults prediction and found that these models achieved good performance. Chen et al [11] performed an empirical study on predicting the number of faults using six regression algorithms and found that the prediction model built with decision tree regression had the highest prediction accuracy in most cases. In another similar study, Rathore et al [12] presented an experimental study to evaluate and compare the other six regression algorithms for the number of faults prediction.…”

A Ranking-Oriented Approach to Cross-Project Software Defect Prediction: An Empirical Study

“…In these CPDP scenarios, the focus of the researches most relevant to the topic of this paper is on predicting or estimating the number of software defects/faults in a given software entity, which could be deemed as a specific problem with predictive analytics [10,21]. In fact, this is not a new field of study.…”

“…However, estimating the defect-proneness of a given set of classes or software modules has limited effect on actual activities in software testing and software maintenance [7,9,10], especially when there is a lack of human resources. That is to say, from a software developer's point of view, a ranking list of defect-prone software entities is definitely more useful than the information about how many the software entities in question are possibly buggy.…”