The relationship between object oriented metrics and software maintenance effort is complex and non-linear. Therefore, there is considerable research interest in development and application of sophisticated techniques which can be used to construct models for predicting software maintenance effort. The aim of this paper is to evaluate and compare the application of different soft computing techniques -Artificial Neural Networks, Fuzzy Inference Systems and Adaptive Neuro-Fuzzy Inference Systems to construct models for prediction of Software Maintenance Effort. The maintenance effort data of two commercial software products is used in this study. The dependent variable in our study is maintenance effort. The independent variables are eight Object Oriented metrics . It is observed that soft computing techniques can be used for constructing accurate models for prediction of software maintenance effort and Adaptive Neuro Fuzzy Inference System technique gives the most accurate model.

The objective of this paper is statistical comparison of modelling methods for software maintainability prediction. The statistical comparison is performed by building software maintainability prediction models using 27 different regression and machine learning based algorithms. For this purpose, software metrics datasets of two different commercial object-oriented systems are used. These systems were developed using an object oriented programming language Ada. These systems are User Interface Management System (UIMS) and Quality Evaluation System (QUES). It is shown that different measures like MMRE, RMSE, Pred(0.25) and Pred(0.30) calculated on predicted values obtained from leave one out (LOO) cross validation produce very divergent results regarding accuracy of modelling methods. Therefore the 27 modelling methods are evaluated on the basis of statistical significance tests. The Friedman test is used to rank various modelling methods in terms of absolute residual error. Six out of the ten top ranked modelling methods are common to both UIMS and QUES. This indicates that modelling methods for software maintainability predicton are solid and scalable. After obtaining ranks, pair wise Wilcoxon Signed rank test is performed. Wilcoxon Sign rank test indicates that the top ranking method in UIMS data set is significantly superior to only four other modelling methods whereas the top tanking method in QUES data set is significantly superior to 11 other modelling methods. The performance of instance based learning algorithms — IBk and Kstar is comparable to modelling methods used in earlier studies.

Machine learning techniques have been earnestly explored by many software engineering researchers. At present state of art, there is no conclusive evidence on the kind of machine learning techniques which are most accurate and efficient for software defect prediction but some recent studies suggest that combining multiple machine learners, that is, ensemble learning, may be a more accurate alternative. This study contributes to software defect prediction literature by systematically evaluating the predictive accuracy of three well known homogeneous ensemble methods -Bagging, Boosting, and Rotation Forest, utilizing fifteen important underlying base learners, by exploiting the data of nine open source objectoriented systems obtained from the PROMISE repository. Results indicate while Bagging and Boosting may result in AUC performance loss, AUC performance improvement results in twelve of the fifteen investigated base learners with Rotation Forest ensemble.

No abstract

No abstract