Replication Kit: On the Feasibility of Automated Classification of Bug and Non-Bug Issues

This is the replication kit for the manuscript "On the Feasibility of Automated Issue Bug and Non-Bug Issues". The replication kit contains the following data.<br> - results_all_issues contains the csv files with the results of the classifiers we trained with all issues for phases 1 and 2 of the experiment.<br> - results_only_bugs contains the csv files with the results of the classifiers we trained with only bugs for phases 1 and 2 of the experiment.<br> - results_unvalidated contains the csv files with the results for phases 3 and 4 of the experiment<br> - The .p files contain pickled python objects with the issue data. The code for creating this pickles is contained in the evaluation.py.<br> - The evaluation.py creates all results that are missing, i.e., will do nothing if there are already CSV files in the folders! <br> - The evaluation.py also generates the .p files, if they are not available. However, this requires the raw data from Herzig et al (2013), Herbold et al (2020), and Ortu et al (2015) which is not included in this replication kit due to the size of several gigabytes.<br> - The EvaluationNotebook contains the code for the statistical analysis.<br>本套件为论文《论自动化区分问题中的缺陷与非缺陷问题的可行性》（"On the Feasibility of Automated Issue Bug and Non-Bug Issues"）的复现工具包。本复现工具包包含如下数据：<br> - results_all_issues：存储实验阶段1与阶段2中，使用全量问题训练分类器所得结果的CSV文件。<br> - results_only_bugs：存储实验阶段1与阶段2中，仅使用缺陷问题训练分类器所得结果的CSV文件。<br> - results_unvalidated：存储实验阶段3与阶段4的结果的CSV文件。<br> - 后缀为.p的文件为存储了问题数据的经pickle序列化的Python对象（pickled Python objects），生成该序列化文件的代码位于evaluation.py中。<br> - evaluation.py 可生成缺失的结果文件：若对应文件夹中已存在CSV文件，则不会执行重复操作。<br> - 若所需的.p文件不存在，evaluation.py 也可自动生成该序列化文件，但该步骤需要用到Herzig等人（2013）、Herbold等人（2020）以及Ortu等人（2015）的原始数据；由于这些数据体积达数吉字节，故未包含在本复现工具包中。<br> - EvaluationNotebook：包含统计分析所用的代码。