Problems with SZZ and Features: An empirical study of the state of practice of defect prediction data collection

TL;DR

This empirical study critically examines the reliability of the SZZ algorithm for defect labeling and the impact of feature sets on defect prediction accuracy, revealing significant issues with label correctness and feature importance.

Contribution

The paper provides a comprehensive empirical analysis of SZZ defect labels and feature sets, highlighting their limitations and implications for defect prediction research.

Findings

Only half of SZZ-labeled bug fixing commits are actual bugs.

Using a six-month window causes many mislabels, with one false defect per correct one.

Small feature sets have a less significant impact on defect prediction accuracy.

Abstract

Context: The SZZ algorithm is the de facto standard for labeling bug fixing commits and finding inducing changes for defect prediction data. Recent research uncovered potential problems in different parts of the SZZ algorithm. Most defect prediction data sets provide only static code metrics as features, while research indicates that other features are also important. Objective: We provide an empirical analysis of the defect labels created with the SZZ algorithm and the impact of commonly used features on results. Method: We used a combination of manual validation and adopted or improved heuristics for the collection of defect data. We conducted an empirical study on 398 releases of 38 Apache projects. Results: We found that only half of the bug fixing commits determined by SZZ are actually bug fixing. If a six-month time frame is used in combination with SZZ to determine which bugs affect a release, one file is incorrectly labeled as defective for every file that is correctly labeled as defective. In addition, two defective files are missed. We also explored the impact of the relatively small set of features that are available in most defect prediction data sets, as there are multiple publications that indicate that, e.g., churn related features are important for defect prediction. We found that the difference of using more features is not significant. Conclusion: Problems with inaccurate defect labels are a severe threat to the validity of the state of the art of defect prediction. Small feature sets seem to be a less severe threat.