ceemaviya/geodrugsai_zim_MLpredict

predict hotspot areas for drug abuse in selected provinces in Zimbabwe using ML # GeoDrugs AI Full Stack System with Maps and Reports Predict hotspot areas for drug abuse in selected provinces in Zimbabwe using ML. Includes FastAPI backend, React frontend, login, dashboard, prediction using trained model variables, hotspot map, time series charts, reports, variables page and logout. Also includes future hotspot forecasting using projected socio-economic factors such as unemployment rate, crime rate, poverty index, treatment access, school dropout and urban/rural setting. ## Backend cd backend python -m pip install -r requirements.txt python -m uvicorn main:app --reload Put your model file here: backend/geodrucgs_model.pkl Default login: - Username: `admin` - Password: `admin123` For local configuration, set these environment variables before starting the backend: - `GEODRUGS_USERNAME` - `GEODRUGS_PASSWORD` - `GEODRUGS_AUTH_TOKEN` - `GEODRUGS_ALLOWED_ORIGINS` ## Frontend cd frontend npm install npm run dev Open localhost Set `VITE_API_URL` if your backend is not running at `127.0.0.1`. If prediction says model not loaded, copy geodrucgs_model.pkl into the backend folder. ## Future hotspot forecasting Use the frontend page: `/future-hotspots` Backend endpoint: POST `127.0.0.1` ## CSV batch prediction Use the Prediction page and upload a `.csv` file. Required columns: `province, district, latitude, longitude, reported_incidents, historical_incident_frequency, poverty_level, unemployment_rate, crime_rate, population_density, health_service_access, urban_rural_class, prior_hotspot_status, education_attainment, neighboring_district_case_density, year, month_number, quarter, is_holiday_season, time_index, incident_growth, incident_growth_rate, incidents_per_density, socioeconomic_pressure_index, service_deprivation_index, crime_density_interaction, hotspot_persistence_index, neighbor_pressure_index` Example file: `sample_predictions.csv` Backend endpoint: POST `127.0.0.1 …