FSCR:A Feature Selection Method for Software Defect Prediction

Abstract

Abstract-Prediction the number of faults in software modules can be more helpful instead of predicting the modules being faulty or non-faulty. Some regression models have been used for predicting the number of faults. However, the software defect data may involve irrelevant and redundant module features, which will degrade the performance of these regression models. To address such issue, this paper proposes a feature selection method based on Feature Spectral Clustering and feature Ranking (FSCR) for the number of software faults prediction. First, FSCR groups the original features with spectral clustering according to the correlation between every two features. Second, FSCR employs ReliefF algorithm to compute the relevance between each feature with respect to the number of faults and selects top p most relevant features from each resulted cluster. We evaluate our proposed method on 6 widely-studied project datasets with four performance metrics. Comparison with five existing feature selection methods demonstrates that FSCR is effective in selecting features for the number of faults prediction.

Few-Shot Learning Based Balanced Distribution Adaptation for Heterogeneous Defect Prediction

“…The researchers used the feature selection method to select shared metrics from the source domain and target domain, and constructed a unified model [6], [2]. Yu et al grouped the original metrics with spectral clustering according to the correlation of metrics, and employed Relief F algorithm to compute the relevance between each metric with respect to the number of faults, and selected the most relevant metrics from each resulted cluster [7]. Subspace learning transforms the source domain and the target domain into the same subspace.…”