Bus Route Overlap & Rationalization Tool (Streamlit)

A Streamlit port of the original Jupyter/ipywidgets notebook. Same analysis logic, same two tabs, same maps, drawing, upload, Excel export, and the full N×N (98×98) network self-overlap matrix — now runnable as a standalone web app instead of a notebook.

Project layout

bus_route_app/
├── app.py            # Streamlit UI: session state, both tabs, results rendering
├── analysis.py        # Pure analysis logic (geometry, overlap, scoring, matrix, I/O)
├── map_utils.py        # Folium map builders (results map + draw-mode input map)
├── requirements.txt
├── data/               # Put your network shapefiles here (see below)
└── README.md

1. Install dependencies

python -m venv venv
source venv/bin/activate        # Windows: venv\Scripts\activate
pip install -r requirements.txt

2. Add your network data

Place these files in the data/ folder (or point the app at any folder via the sidebar "Data folder" field):

• Official_Bus_Routes_98.shp (+ .dbf, .shx, .prj, …) — needs a Route_ID field, and optionally Operator.
• Official_Bus_Stops_98.shp — needs a Name field.
• Bus_Corridors.shp — needs a Name field.
• 98_Overlap_Matrix.xlsx — optional, precomputed matrix (not required to run the app; the app can compute this itself from Tab 2).

Any CRS is fine — the app reprojects everything to EPSG:32646 internally for length/overlap math and to EPSG:4326 for the map.

3. Run

streamlit run app.py

Then open the local URL Streamlit prints (usually http://localhost:8501).

What each tab does

🆕 New / Custom Route

• Enter Route ID / Origin / Destination.
• Provide the route geometry either by uploading a file (GeoJSON, GeoPackage, or a zipped Shapefile) or by drawing a polyline directly on the map, then clicking Use drawn route.
• Pick a matching tolerance and click ANALYZE.
• Get the results map, overlap table, conflict score, recommendation, the unique affected-bus-stops table (Name / Latitude / Longitude, EPSG:4326, de-duplicated), affected corridors, and an Excel export.

🔁 Existing Network Self-Overlap

• Pick any of the existing routes and compare it against the rest of the network (same result view as above).
• Or click Compute full N×N overlap matrix to run every route against every other route, with a live progress bar, a ranked table of the top overlapping pairs, and an Excel export of the full matrix + pairs table.

Notes on the port from the notebook

• ipywidgets → Streamlit widgets (st.text_input, st.selectbox, st.file_uploader, st.button, session state for everything the notebook used module-level globals for).
• ipyleaflet DrawControl → folium.plugins.Draw, rendered with streamlit_folium.st_folium, reading the drawn line back from last_active_drawing.
• All core geometry/overlap/scoring functions in analysis.py are unchanged in behavior from the notebook (same tolerance logic, same 0.55/0.25/0.10/0.10 conflict-score weighting, same HIGH/MEDIUM/LOW/MINIMAL severity thresholds).
• Excel export now returns in-memory bytes for st.download_button instead of writing to a fixed Windows path.