mbuguasamuelwambui/mlfc_mini-project

Nationwide Mapping of Power Infrastructure and Environmental Features in Kenya Using Open Geospatial Data # ⚡ MLFC Mini-Project: Mapping Power Infrastructure in Kenya This project explores **where and how power stations can be strategically set up across Kenya** by combining **open geospatial data** (OSM, GADM, DEM, climate/weather datasets) with *machine learning*. It is built on the *Fynesse framework*, which structures data work into three modular steps: - **Access** → Collect & preprocess open datasets (counties, power stations, OSM rivers/roads, etc.) - **Assess** → Validate & visualize environmental and infrastructure features - **Address** → Model and generate insights, e.g., predicting optimal power station sites with Gaussian Processes The ultimate goal is to provide a reproducible, data-driven foundation for **sustainable energy planning, infrastructure expansion, and environmental stewardship** in Kenya. ## 🧭 Objectives - Map Kenya’s power stations and surrounding environmental features - Query and clip OSM features (rivers, lakes, forests, roads, grids, etc.) at the county level - Compute distance-based features for each power station (e.g., distance to river, distance to grid) - Train a Gaussian Process Classifier to model probability of power station presence given environmental conditions - Build a scalable pipeline with outputs saved to CSV/GeoJSON for reuse ## Requirements To run the notebook, you’ll need the following dependencies: ```bash pip install geopandas matplotlib contextily osmnx scikit-learn fiona shapely ``` ## 🧱 Fynesse Framework | Module | Purpose | | -------------------- | -------------------------------------------------- | | `fynesse/access.py` | Download & preprocess open datasets | | `fynesse/assess.py` | Data validation, visualization, feature extraction | | `fynesse/address.py` | Modeling & answering key research questions | ## Quick Start ### Prerequisites - Python 3.9 or higher - Poetry (install via `curl -sSL insta …