Replication Package for the paper: "Predicting Design Impactful Changes in Modern Code Review: A Large-Scale Empirical Study"

This is the replication package for the paper: "Predicting Design Impactful Changes in Modern Code Review: A Large-Scale Empirical Study", published at the 18th International Conference on Mining Software Repositories (MSR'21).

It contains all the preliminary and final results of our empirical methodology as follows.
 

• description_and_detection_mechanisms.zip: This file contains detailed descriptions for all types of symptoms.
• extracted_degradation_symptoms_per_revision.zip: This file contains the raw data of types of symptoms extracted for all revisions.
• design_impactful_changes_per_revision.zip: This file contains the raw data of design impactful changes per revision and system.
• extracted_features_per_revision.zip: This file contains the raw data of extracted features (both social and technical features) per revision and system.
• generated_ml_models.zip: This file contains the complete set of generated machine learning models grouped by feature set (social only, technical only, and social + technical together) and symptom category (fine-grained smells and coarse-grained smells).
• performance_ml_models.zip: This file contains the complete description of the performance evaluation of the machine learning models considering the different combinations of feature sets and symptom categories (fine-grained smells and coarse-grained smells).
• features_importance.zip: This file contains the complete set of features importance grouped by different rankings.
• features_statistics.zip: This file contains the complete set of descriptive analysis of extracted features (social and technical) per system.
• friedman_test_with_post_hoc_statistic.zip: This file contains the results, scripts, and visualizations of the Friedman with post hoc test employed in RQ2 and RQ3.
• wilcoxon_script.zip: This file contains the script used in RQ1.
• features_description.zip: This file contains the complete description of extracted features (social and technical).
• hyperparametrization.zip: This file contains the hyper parameterization used in each ml model grouped by feature set, system, and smell granularity.

Feel free to use any part of this replication package in your study, please cite as:

Anderson Uchôa, Caio Barbosa, Daniel Coutinho, Willian Oizumi, Wesley K. G. Assunção, Silvia Regina Vergilio, Juliana Alves Pereira, Anderson Oliveira, Alessandro Garcia. Predicting Design Impactful Changes in Modern Code Review: A Large-Scale Empirical Study. Proceedings of the 18th International Conference on Mining Software Repositories (MSR),  Madrid, Spain, May 2021.