MPR_Model provides a model training and prediction framework for water quality evaluation based on multivariate polynomial regression and machine learning approaches. The solution supports standard indicators and generates Water Quality Index (WQI/WQI5) estimates.
- Support for science-backed indicators: DO, BOD, NH3-N, EC, SS
- Model variants maintained in
models/(e.g., LR, RF, SVM, XGBoost, LightGBM, MPR) - REST API implementation under
src/api.pyfor inference and data ingestion - Training scripts in
src/trainingfor reproducibility on various datasets - Example data in
data/for quick evaluation and comparison
- Python 3.8 or later
- Virtual environment manager (recommended):
venvorconda
git clone https://github.com/KageRyo/MPR_Model.git
cd MPR_Model
python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txtmain.pydefaults to listening on0.0.0.0:8000.- Update host/port in
main.pyor extend configuration handling insrc/api.pyas needed.
python3 main.pypython3 src/training/mainLGBMVer.1.0.pyPOST /predict: accepts JSON or CSV payload for one-to-many inferenceGET /status: health check endpoint
Refer to src/api.py for exact endpoint paths and data schemas.
Accepted CSV shape: | DO | BOD | NH3N | EC | SS | Rows should contain numeric values.
data/: training and validation datasetsmodels/: persisted model artifactssrc/: API and training implementationsmain.py: service entrypoint
- Include linting (
flake8,pylint) in your development process. - Add tests for new behavior under
src/test.
- Fork the repository
- Create a feature branch
- Add tests and documentation updates
- Submit a pull request with verification steps
Apache License Version 2.0. See LICENSE.
- Chien-Hsun Chang (KageRyo)
- Kuo-Wei Wu (RRAaru)
