Meliodus254/BEE-OPTIMUM-PLACEMENT-PROJECT

Data science project for identifying optimal beehive placement in Kenya using climate, vegetation, and environmental data. # 🐝 Bee Optimal Placement Project ## Machine-Learning Suitability Analysis – Kenya --- ## Project overview This project uses a multi-criteria ML pipeline to predict the most suitable locations for honeybee hive placement across Kenya. **Data inputs:** | Layer | Files | Purpose | |-------|-------|---------| | Temperature | `climatology-tas-…nc` | ERA5 monthly mean (°C) | | Wind speed | `climatology-sfcwind-…nc` | ERA5 surface wind (m/s) | | Precipitation | `natvar-pr-…nc` | ERA5 seasonal rain (mm/yr) | | Solar radiation | `natvar-rsds-…nc` | ERA5 shortwave down (W/m²) | | Water bodies | `gis_osm_water_a_free_1.shp` + `gis_osm_waterways_free_1.shp` | Proximity to water | | Land cover | `gis_osm_landuse_a_free_1.shp` + `gis_osm_natural_a_free_1.shp` | Vegetation quality | | Roads | `gis_osm_roads_free_1.shp` | Accessibility | | Buildings | `gis_osm_buildings_a_free_1.shp` | Urban buffer | **ML pipeline:** 1. 0.1° (~11 km) grid sampled over Kenya 2. Climate variables interpolated via `scipy.RegularGridInterpolator` 3. Spatial distances computed with `scipy.cKDTree` (fast KD-tree) 4. Physics-based suitability score computed as weighted sum 5. Random Forest (300 trees) + Gradient Boosting (200 iters) trained 6. Ensemble prediction applied to full grid 7. Two interactive Folium HTML maps generated --- ## Quick start ### 1. Install dependencies ```bash pip install -r requirements.txt ``` Tested on Python 3.10/3.11. If you hit conflicts, use a virtual env: ```bash python -m venv bee_env bee_env\Scripts\activate # Windows source bee_env/bin/activate # Linux/Mac pip install -r requirements.txt ``` ### 2. Verify your data directory Open `config.py` and check: ```python DATA_DIR = r"C:\Users\ZUPLO\Desktop\BEE PROJECT" ``` Change this if your data is elsewhere. ### 3. Run the pipeline ```bash cd "C:\Users\ZUPLO\Desktop\BEE PROJECT" python main.py ``` This will take **15–40 minutes** the first time (the road and building shapefiles are large – 1. …