eowusu1234/ghana-hospital-map

This project intends to display all hospital locations in Ghana # UNDP Ghana Disease Prediction — Malaria / Diarrhoea / Cholera Risk Model Scaffold project for predicting community-level disease risk (as a continuous probability, 0–1) from environmental, WASH (water/sanitation/hygiene), and demographic determinants — built for the Ada East (+ Tamale / Wa East) facility and drone data collection effort. Currently trained on **synthetic data**. Swap in real facility/drone-derived data by matching the column schema in `config.py` — no other code changes needed. ## Project structure ``` disease_prediction/ ├── config.py # feature groups, sites, targets, paths ├── data/ │ ├── synthetic_generator.py # generates synthetic communities.csv │ └── synthetic_communities.csv # generated raw data ├── features/ │ └── feature_engineering.py # derived features + categorical encoding ├── models/ │ ├── train.py # trains one regressor per disease │ └── predict.py # scores new communities ├── outputs/ │ ├── models/ # saved .joblib models + training_metrics.json │ └── risk_predictions.csv # latest scored output └── requirements.txt ``` ## Features used **Environmental:** temperature, rainfall, humidity, flooding index, drought index, elevation, soil moisture, distance to water, water quality index. **WASH:** clean water access, water source type, distance to source, water treatment rate, handwashing facilities, soap availability, toilet availability, open defecation rate, sewer coverage, waste disposal. **Demographic:** population density, household size, % under-five, median age, education index, income index, urban/rural, migration/displacement flag. Each disease model also gets **derived composite features** (see `features/feature_engineering.py`): - `water_risk_composite`, `sanitation_composite`, `hygiene_composite` - `temp_suitability_malaria` (peaks near 27°C, the *Anopheles*-friendly range) - `social_vulnerability_index`, `crowding_i …