yzh-han/water-pump-status-prediction

This is a machine learning project that focuses on water pump status prediction in Tanzania using various ML algorithms, custom preprocessing pipelines, and Optuna-based hyperparameter optimization. # Water Pump Status Prediction This is a machine learning project that focuses on water pump status prediction in Tanzania using various ML algorithms, custom preprocessing pipelines, and Optuna-based hyperparameter optimization. This is a coursework for Artificial Intelligence Practice at University of St Andrews. ## Project Structure ``` bash repo/ ├── my_classifier/ # Main package directory │ ├── analysis/ # Analysis notebooks and results │ │ ├── hpo_300.db # Hyperparameter optimization results │ │ ├── model_train_scores.csv # Cross validation results │ │ ├── part1_analysis.ipynb # Part 1 analysis notebook │ │ ├── part2_analysis.ipynb # Part 2 analysis notebook │ | ├── requirements.txt # Run analysis notebook dependencies │ │ └── row_data_analysis.ipynb # Raw data analysis notebook │ ├── hpo.py # Hyperparameter optimization mplementation │ ├── my_classifier.py # Main classifier implementation │ ├── my_transformers/ # Custom transformers │ │ └── my_transformers.py # Transformer implementations │ ├── requirements.txt # Package dependencies │ └── test_myCLF.py # Model testing script ├── requirements.txt # Project dependencies └── train-and-predict.py # Main training script ``` ## Setup Install the required dependencies: ```bash pip install -r requirements.txt ``` ### Requirements Main dependencies include: - scikit-learn - pandas - numpy - optuna See at `requirements.txt` ## Usage ### Part 1: Training and Prediction Use the `train-and-predict.py` script to train a model and generate predictions: ``` bash python train-and-predict.py # e.g. # python train-and-predict.py ../data/training-input.csv ../data/training-labels.csv ../data/test-input.csv StandardScaler OrdinalEncoder HistGradientBoostingClassifier ../ …