Replication Package: Code Quality Evaluation Using Machine Learning and SonarQube

Replication package for the undergraduate thesis "Evaluación de calidad de código usando Machine Learning y SonarQube" presented at Universidad de Guayaquil, Faculty of Mathematical and Physical Sciences, Software Engineering Program. This package contains:- Python scripts (01–05) implementing the complete ML pipeline- Processed dataset (N = 2,308 records, balanced)- Trained model (XGBoost C3)- Comparative results (Random Forest vs XGBoost) Pipeline overview:1. Data extraction from Technical Debt Dataset V2.0 (Jira + SonarQube)2. Preprocessing: temporal normalization, merge_asof, deduplication (62.08% removed)3. Feature engineering: 5 predictors, Gaussian Gap P15/P85, RobustScaler4. Training: 6 configurations per algorithm (RF and XGBoost), 80/20 stratified split5. Evaluation: Accuracy, MCC, F1-Score, AUC-ROC, 5-fold stratified CV, Wilcoxon signed-rank test Requirements: Python 3.11, scikit-learn, xgboost, scipy, pandas, numpy, matplotlibExternal dependency: Technical Debt Dataset V2.0 (td_V2.db) — available at https://github.com/clowee/The-Technical-Debt-Dataset