sabouziane-coder/Moulouya_Water_Quality

Dataset of 66 samples of water physiochemical parameters from the Moulouya River basin, Morocco # Moulouya River Water Quality — Pollution Index (PI) Prediction This repository provides the data and Python scripts associated with the study **“An Ensemble Learning Approach for Water Quality Assessment and Prediction Using the Pollution Index: A Case Study of the Moulouya River Basin, Morocco.”** The repository includes preprocessing routines, Pollution Index (PI) calculation, baseline and ensemble model training, model evaluation and robustness analysis, and the figures used in the manuscript. --- ## Repository Structure ```text . ├── data/ │ ├── raw/ │ │ ├── README.md │ │ └── standards.csv │ └── processed/ │ └── Final_Pollution_Index_Results.csv ├── src/ │ ├── preprocess.py │ ├── calc_pi.py │ ├── training_models_pi.py │ └── plot_ensemble.py ├── requirements.txt ├── README.md └── .gitignore ``` ## Directory Description ### `data/raw/` Contains reference tables and metadata used for Pollution Index (PI) computation. Due to data ownership restrictions, original raw measurement files are not publicly shared. The `standards.csv` file reconstructs the thresholds and weights required for PI calculation. ### `data/processed/` Contains the processed and analysis-ready dataset used for model training and evaluation. `Final_Pollution_Index_Results.csv` includes all physicochemical parameters and the computed PI values. ### `src/` Python scripts implementing the full modeling pipeline: - `preprocess.py`: data cleaning, normalization, and preparation - `calc_pi.py`: Pollution Index (PI) computation - `training_models_pi.py`: model training, cross-validation, evaluation, and uncertainty analysis - `plot_ensemble.py`: generation of performance, diagnostic, and comparison figures ### `requirements.txt` Lists all Python dependencies with fixed versions to ensure full reproducibility.