- Updated comments and code formatting in `3-td_ggplot2 - enonce.Rmd` for clarity.
- Enhanced code structure in `4-td_graphiques - enonce.Rmd` by organizing options and library calls.
- Replaced pipe operator `%>%` with `|>` in `Code_Lec3.Rmd` for consistency with modern R syntax.
- Cleaned up commented-out code and ensured consistent spacing in ggplot calls.
- Created a new Excel file: `departements-francais.xlsx` for data storage.
- Added a CSS file: `style.css` with custom styles for various mathematical environments including boxes for lemmas, theorems, definitions, and more, complete with automatic numbering.
- Initialized R project file: `tp2.Rproj` with default settings for workspace management and LaTeX integration.
- Added CatBoost version 1.2.8 to the project dependencies in pyproject.toml.
- Updated uv.lock to include CatBoost and its dependencies, along with the necessary wheel files.
- Included Graphviz version 0.21 in the lock file as a dependency for CatBoost.
- Updated execution counts for various code cells to maintain consistency.
- Changed the model from RandomForestClassifier to GradientBoostingClassifier.
- Modified hyperparameter grid for GridSearchCV to include learning_rate and adjusted n_estimators.
- Added stratification to train-test split for better representation of classes.
- Corrected scoring parameter in GridSearchCV to use a valid metric.
- Updated output messages to reflect changes in model evaluation metrics.
- Created a new Jupyter notebook: 2025_M2_ISF_TP_4.ipynb for supervised machine learning exercises, including data preparation, model building, and performance analysis.
- Added 'imblearn' as a dependency in pyproject.toml to support handling imbalanced datasets.
- Updated uv.lock to include the 'imbalanced-learn' package and its dependencies.
- Changed execution_count from 3 to null for a cleaner notebook state.
- Simplified the normality test logic by using a conditional expression to determine the p-value calculation, improving code readability.
- Updated execution counts for multiple code cells to maintain consistency.
- Removed redundant imports and organized import statements.
- Improved formatting for better readability in train-test split section.
- Added markdown explanations for model performance metrics (MAE, RMSE).
- Enhanced cross-validation training loop with detailed output for each fold's metrics.
