MainaE677/Matatu-Analysis-Kenya

Analysis of transport(PSVs) routes in Kenya # Nairobi Matatu Stop Coverage Analysis A self-directed geospatial data analytics project analyzing coverage of Nairobi's informal matatu transit network, using open GTFS transit data and administrative ward boundaries. ## Question Where does Nairobi's informal matatu transit network provide strong coverage, and where are the coverage gaps — and how does that pattern relate to ward population and income level? ## Data Sources - **Digital Matatus GTFS Feed (2019)** — a University of Nairobi / MIT / Columbia collaboration mapping 140+ matatu routes and 4,000+ stops across Nairobi - **Kenya Administrative Wards** (Humanitarian Data Exchange) — ward-level boundary polygons with 2009 census population, filtered to Nairobi County's 85 wards Raw data files are not included in this repository due to size — see Reproducing this analysis below for download links. ## Method 1. Loaded GTFS stop, route, and trip data with **pandas**; converted stop coordinates into point geometries with **geopandas** 2. Loaded Nairobi ward boundary polygons and performed a **spatial join** to assign each matatu stop to its containing ward 3. Calculated three complementary coverage metrics per ward: - Raw stop count - Stops per 10,000 residents (using 2009 census population) - % of ward land area within a 500m walking buffer of the nearest stop (reprojected to UTM Zone 37S for accurate metric distance) 4. Built a labeled choropleth map in matplotlib visualizing ward-level coverage ## Key Findings 1. **Coverage is highly uneven**: only 39.7% of Nairobi's land area falls within a 500m walk of a formally-mapped matatu stop. 2. **Three independent metrics converge on the same underserved wards**: Kwa Reuben, Lindi, Huruma, Kware, Dandora Area IV, and Laini Saba rank lowest by raw stop count, stops per capita, *and* % area covered. All are lower-income or informal-settlement areas. The best-served wards (Kileleshwa, Karen, Kilimani, Parklands/Highridge) are higher-income, formally-planned ne …