Bengaluru Mobility & Last-Mile Planning Dashboard

Project Overview

This project builds a public-transport & last-mile accessibility dashboard for Bengaluru.
It aggregates BMTC bus stops, metro stations, and BBMP zones, then computes planning KPIs like access coverage, FLGI (first/last-mile gap index), bus stop density, and nearest-metro distance.

This dashboard helps city planners pinpoint where public transport access is weak, prioritize high-impact investments (new bus stops, feeder routes, sidewalks, cycle links), and track shifts in mode share over time. By making it easier to use buses and metro—and improving first/last-mile connectivity—it shifts trips away from private vehicles, cutting congestion, air pollution, and CO₂ emissions across the city.

The final deliverable is a Streamlit app that lets planners:

• Compare zones on supply and access.
• Explore gaps (where to add stops/feeder/footpaths).
• Inspect bus stop ↔ metro station distances and download the filtered data.
• Track citywide mode share KPIs (PT / NMT / Private) over time.

This Project was created as a contribution at the Bengaluru Climate Hackathon, hosted by WRI India, BBMP, OpenCity, BCAP and partners.

---

Features in Streamlit App

• KPIs & Trend
  Citywide PT/NMT/Private shares by year, YoY deltas, and trend lines. (Data from your CSV.)
• Supply & Access (by Zone)
  Zone selector → grid heatmap of Access, FLGI, Bus stop density, Nearest metro, Stations ≤800 m, plus histogram/ECDF/rank charts and Bottom-10 table.
• Stop ↔ Station Distance Explorer
  Shows nearest metro station for every BMTC stop; filter by distance slider and download CSV.
• Auto cache-busting & Refresh
  Caches keyed to file modification time; sidebar button to clear cache.
• Downloadables
  Filtered stop–station pairs as CSV; zone tables exportable from the UI.

---

Implementation Steps (Phase 1 → Phase 7)

Phase 1 — Base layers

• Fetch Bengaluru boundary via OSM (Nominatim through OSMnx).
• Save: processed/bengaluru_boundary.geojson

Phase 2 — Supply data

• Metro stations from Overpass (OSM); filter by metro keywords.
• BMTC stops from community GTFS-derived GeoJSON.
• Save:
  • processed/stations_overpass.geojson
  • processed/bmtc_stops.geojson

Phase 3 — Precompute distances (stop ↔ nearest station)

• For each BMTC stop, compute Haversine distance to all stations; keep the minimum.
• Save:
  • processed/stop_station_nearest.csv (one row per stop)
  • processed/station_nearby_stops.json (station → stops within a radius)

Phase 4 — Access & FLGI (grid metrics)

• Build a square grid (e.g., 2 × 2 km) clipped to the city.
• Access coverage: % of each cell within 500 m of any bus stop (union of buffers).
• FLGI proxy: 1 − Access.
• Also compute bus stop count/density, nearest metro (m) from cell centroid, stations ≤800 m.

Phase 5 — Zones/Wards

• Download BBMP zones/wards (OpenCity), normalize to zone_id, save:
  • processed/zones.geojson
• Attach zone_id to grid cells via centroid-in-polygon.

Phase 6 — KPIs data (you provide)

• Create processed/modal_share_city.csv with columns:
  year, pt_share_pct, nmt_share_pct, private_share_pct
• (Optional) meta/targets.yaml for PT/NMT targets shown in the header.

Phase 7 — Streamlit dashboard

• App file: notebooks/app_all_in_one.py
• Tabs: KPIs & Trend • Supply & Access (by Zone) • Distance Explorer
• Sidebar: grid size, Refresh data (clear cache).

---

How to Run Locally

most of the work is already done in this (if you want to start it from scratch, follow this steps)

1) Create environment & install Python deps

python -m venv .venv
# macOS/Linux
source .venv/bin/activate
# Windows
# .venv\Scripts\activate

python -m pip install --upgrade pip
pip install streamlit pandas numpy altair pydeck geopandas shapely pyogrio requests fastkml osmnx networkx geopy

2) Build transit layers & distances (Node)

# Metro stations (OSM) + BMTC stops + nearest-station pairs
node scripts/build_bmtc_lastmile.js

3) Generate zones/wards (required)

# Optional: change “nearby stops” radius for station_nearby_stops.json
LASTMILE_RADIUS=1000 node scripts/build_bmtc_lastmile.js

python scripts/make_zones_file.py
# writes processed/zones.geojson with zone_id

4) Provide KPIs CSV (required)

year,pt_share_pct,nmt_share_pct,private_share_pct
2019,41.0,18.5,40.5
2020,41.2,18.6,40.2
...
2025,48.0,18.5,33.5

5) Run the app

.venv/bin/streamlit run notebooks/app_all_in_one.py

Metric Definitions

• Access (coverage): share of a grid cell within 500 m of any bus stop (union of buffers).
• FLGI (proxy): 1 − Access → fraction not covered (higher = worse last-mile).
• Bus stop density: stops per km² in the cell.
• Nearest metro (m): straight-line distance from cell centroid to closest station.
• Stations ≤800 m: number of stations within 800 m of centroid.
• YoY (pp): current year % − previous year % (percentage points).

Data Source

• OpenStreetMap / Overpass API — metro stations & fallback bus stops https://www.openstreetmap.org • https://overpass-api.de
• BMTC GTFS-derived GeoJSON (community) — bus stops https://github.com/Vonter/bmtc-gtfs (uses geojson/stops.geojson)
• Bengaluru boundary (Nominatim via OSMnx) https://nominatim.openstreetmap.org • https://github.com/gboeing/osmnx
• BBMP Zones/Wards (OpenCity) https://data.opencity.in/ (KML downloads; converted to GeoJSON by our script) © OpenStreetMap contributors (ODbL). Check OpenCity pages for dataset licenses.

Team Behind the Work

• Advik R Holalu
• Akshath S Bhandiwad
• Joselyn Riana Manoj
• K Jaidev Shankar Reddy
• Sanjan Dileep