What really changes when developers intend to improve their source code: A commit-level study of static metric value and static analysis warning changes

<p>This is the dataset for the publication &quot;What really changes when developers intend to improve their source code: A commit-level study of static metric value and static analysis warning changes&quot;.</p> <p>It contains a random sample of 2533 commits from 54 Java Apache open source projects classified by two researchers into perfective, corrective and other changes (manual_labels.csv).&nbsp; Moreover, we include static source code metrics and static analysis warnings for the 2533 changes in al_changes_gt.csv.gz.</p> <p>In addition, we include the full dataset of 125482 commits in all_changes_sebert.csv.gz with all metrics and automatic labels for every commit that was not manually labeled. The automatic labels were provided by a fine-tuned transformer model (BERT) pre-trained exclusively on software engineering data.</p> <p>We also provide the fine tuned version of the pre-trained model in seBERT_fine_tuned_commit_intent.tar.gz as well as a Snapshot of the SmartSHARK MongoDB database used in gathering the raw data in smartshark_emse.agz.</p> <p>The model can be tested live on the <a href="https://user.informatik.uni-goettingen.de/~trautsch2/emse_2021/commit_intent.html">website</a> accompanying the publication.</p>