ALTRA: Cross-Project Software Defect Prediction via Active Learning and Tradaboost

Abstract

Cross-project defect prediction (CPDP) methods can be used when the target project is a new project or lacks enough labeled program modules. In these new target projects, we can easily extract and then measure these modules with software measurement tools. However, labeling these program modules is time-consuming, error-prone and requires professional domain knowledge. Moreover, directly using labeled modules in the other projects (i.e., the source projects) can not achieve satisfactory performance due to the large data distribution difference in most cases. In this article, to our best knowledge, we are the first to propose a novel method ALTRA, which can utilize both active learning and TrAdaBoost to alleviate this issue. In particular, we firstly use Burak filter to select similar labeled modules from the source project after analyzing the unlabeled modules in the target project. Then we use active learning to choose representative unlabeled modules from the target project and ask experts to label the type (i.e., defective or non-defective) of these modules. Later, we use TrAdaBoost to determine the weights of labeled modules in the source project and the target project, and then construct the model via weighted support vector machine. After selecting a small number of modules (i.e., only 5% modules) in the target project, we terminate the method ALTRA and return the final constructed model. To show the effectiveness of our proposed method ALTRA, we choose 10 large-scale open-source projects from different application domains. In terms of both F1 and AUC performance indicators, we find ALTRA can perform significantly better than seven state-of-the-art CPDP baselines. Moreover, we also show that the usage of Burak filter, the uncertainty active learning strategy, the class imbalanced learning method and TrAdaBoost are competitive in our proposed method ALTRA.

Cross-Project Defect Prediction Based on Two-Phase Feature Importance Amplification

“…Random Forest is an integrated learning pattern recognition method [ 19 , 38 ]. It has been demonstrated that Random Forest has good performance in CPDP [ 27 , 39 , 40 ] due to its high tolerance to outliers and noise. Random Forest is also less prone to fitting characteristics [ 18 ].…”

“…Yuan et al [ 27 ] used TrAdaBoost to determine weights for samples based on Burak filter and used weighted support vector machines to build the model to improve the CPDP model. Chao et al [ 28 ] proposed a two-phase CPDP method called TPTL, using a source project estimator to select source projects with similar data distribution as the target project and using two improved TCA + to construct models for prediction.…”

“…Various methods have been proposed by researchers for the CPDP problem, such as the classical methods TCA [ 23 ], Peters filtering method [ 22 ], Burak filtering method [ 21 ], and the newer method ALTRA [ 27 ]. We seek to analyse how the performance obtained by TFIA differs from these methods.…”

Defect Prediction Using Akaike and Bayesian Information Criterion

“…According to their study EMBLEM improved P opt 20 and G-scores performance. Yuan et al [8] worked on data labeling in cross-project defect prediction. The proposed method is called ALTRA which uses active learning and TrAdaBoost (Transfer learning Adaboost).…”

Joint Domain Adaption and Pseudo-Labeling for Cross-Project Defect Prediction

“…Zhu et al [7] proposed a method named CooBA based on adversarial transfer learning, which obtains information from cross-project bug reports and focuses only on common features across projects, and CooBA incorporates adversarial learning to ensure that the common information can be extracted effectively. Yuan et al [8] presented an active learning and tradaboost (ALTRA) method. It firstly selects some instances from the source project, which have the most similar distributions to that of the target project, and uses a iterative way in the target instances until the specific number of instances are selected.…”