Abstract:Abstract Effort‐Aware Defect Prediction (EADP) methods sort software modules based on the defect density and guide the testing team to inspect the modules with high defect density first. Previous studies indicated that some feature selection methods could improve the performance of Classification‐Based Defect Prediction (CBDP) models, and the Correlation‐based feature subset selection method with the Best First strategy (CorBF) performed the best. However, the practical benefits of feature selection methods on EADP performance are still unknown, and blindly employing the best‐performing CorBF method in CBDP to pre‐process the defect datasets may not improve the performance of EADP models but possibly result in performance degradation. To assess the impact of the feature selection techniques on EADP, a total of 24 feature selection methods with 10 classifiers embedded in a state‐of‐the‐art EADP model (CBS+) on the 41 PROMISE defect datasets were examined. We employ six evaluation metrics to assess the performance of EADP models comprehensively. The results show that (1) The impact of the feature selection methods varies in classifiers and datasets. (2) The four wrapper‐based feature subset selection methods with forwards search, that is, AdaBoost with Forwards Search, Deep Forest with Forwards Search, Random Forest with Forwards Search, and XGBoost with Forwards Search (XGBF) are better than other methods across the studied classifiers and the used datasets. And XGBF with XGBoost as the embedded classifier in CBS+ performs the best on the datasets. (3) The best‐performing CorBF method in CBDP does not perform well on the EADP task. (4) The selected features vary with different feature selection methods and different datasets, and the features noc (number of children), ic (inheritance coupling), cbo (coupling between object classes), and cbm (coupling between methods) are frequently selected by the four wrapper‐based feature subset selection methods with forwards search. (5) Using AdaBoost, deep forest, random forest, and XGBoost as the base classifiers embedded in CBS+ can achieve the best performance. In summary, we recommend the software testing team should employ XGBF with XGBoost as the embedded classifier in CBS+ to enhance the EADP performance.

The Role of Attribute Ranker Using Classification for Software Defect-Prone Data-sets Model: an Empirical Comparative Study

An empirical analysis of feature selection techniques for Software Defect Prediction

Combined Classifier for Cross-Project Defect Prediction: an Extended Empirical Study.

An Empirical Study on the Equivalence and Stability of Feature Selection for Noisy Software Defect Data

ELM and KELM based software defect prediction using feature selection techniques

Empirical Evaluation Of Classifiers For Software Risk Management

An Empirical Study of Learning to Rank Techniques for Effort-Aware Defect Prediction

Classifier Evaluation for Software Defect Prediction

A Novel Multiple Ensemble Learning Models Based on Different Datasets for Software Defect Prediction

The Impact of Feature Selection on Defect Prediction Performance: an Empirical Comparison

Software defect prediction using learning to rank approach

A feature selection model for software defect prediction using binary Rao optimization algorithm

The Impact of Feature Importance Methods on the Interpretation of Defect Classifiers

Principal component based support vector machine (PC-SVM): a hybrid technique for software defect detection

The Impact of Feature Selection Techniques on Effort-Aware Defect Prediction: an Empirical Study.

Feature Importance in the Context of Traditional and Just-In-Time Software Defect Prediction Models

Classifying feature description for software defect prediction

Software Defect Prediction Analysis Using Machine Learning Techniques

A New Attribute Selection Method Based on Maximal Information Coefficient and Automatic Clustering.

Effective Attribute Selection for Multi-dimensional Root Cause Analysis.

Software Defect Prediction Model Using AdaBoost based Random Forest Technique