The Impact of Mislabeled Changes by SZZ on Just-in-Time Defect Prediction (Code and Data)

This repository provides the datasets and code for reproducing our paper - The Impact of Mislabeled Changes by SZZ on Just-in-Time Defect Prediction.

Data Labelling

Our data is labeled using Daniel's implementaiton of B-SZZ, AG-SZZ, MA-SZZ and RA-SZZ. See the following repositories:

• Repository for B-SZZ, AG-SZZ, MA-SZZ: https://github.com/danielcalencar/ma-szz

• Repository for RA-SZZ: https://github.com/danielcalencar/raszzprime

• the file labeled_data.zip is calculated for all the apache projects using b-szz, ag-szz, ma-szz and ra-szz. All the id are the CVS id of the commit. The CSV id can be retrieved from the commit message of each commit.

How to run

• Clone the code to a directory. Running our code needs the path of the directory.
• Modify a line in the file code/calculation/classification_importance.R, the line is as follows:

# Specify the DIRECTORY path storing the code of this repository
DIR_PATH = "?" 

• Create the following directories in data_results/: importance, oneway, results_imbalance, results_balance.
• Run the script code/calculation/classification_importance.R

Results

The results will be stored in data_results/importance/, data_results/oneway/, data_results/results_imbalance/, data_results/results_balance/ after running the script. We store the performance scores (e.g., auc) in each of the 1,000 bootstrap iterations into csv files.

Citation

If you find our code useful for your research, please cite:

@article{fan2019impact, 
author={Y. {Fan} and X. {Xia} and D. {Alencar da Costa} and D. {Lo} and A. E. {Hassan} and S. {Li}}, 
journal={IEEE Transactions on Software Engineering}, 
title={The Impact of Changes Mislabeled by SZZ on Just-in-Time Defect Prediction}, 
year={2019},
doi={10.1109/TSE.2019.2929761}
}