Working with Maps: Extracting Features for Traffic Crash...
summary
Feature engineering pipeline extracts city road networks with OpenStreetMap via osmnx, normalizes projections with pyproj and geopandas, indexes geometry with rtree, and maps point data to a custom road segment abstraction (20 m buffers merged). Comparison finds 260 vs 175 signals between OSM and municipal Cambridge data, with 17 municipal-only and 41 OSM-only; segment matching to Boston municipal data hits low‑to‑mid‑90s accuracy. Resulting model predicts and visualizes crash risk per segment and is usable for city-scale analysis.
AI-written summary · about
chapters (23)
- 0:00Working with maps for traffic crash analysis
- 1:26OpenStreetMap for feature extraction
- 3:37Downloading road network with OSMnx
- 5:24Simplified vs unsimplified road networks
- 6:08OpenStreetMap tags for features
- 7:57OSM vs municipal datasets
- 9:21OSM nodes and intersections limitations
- 10:03Creating road segments for analysis
- 11:55Standardizing geographic data
- 13:37Reprojecting data with GeoPandas
- 15:26Visualizing dataset mismatches
- 16:10Mapping points to road segments
- 17:34Analyzing mismatched traffic signals
- 19:49Lessons from data inconsistencies
- 20:30Data limitations and geographic issues
- 21:36Adding point-based features
- 22:17Combining multiple maps
- 23:40Matching segments between maps
- 25:05Insight Lane crash risk model
- 25:47Avoiding equity pitfalls
- 27:52Displaying crash risk model
- 28:33Getting involved with Insight Lane
- 29:36Q&A
related talks
13mReimagining Public Transport: An OSM-based Street Supply Tool
18mWhere the Sidewalk Ends Data Driven Urban Safety
22mUnderstanding our cities with FOSS4G
41mLarge Road-Network analysis with GraphX
50mMaking use of OpenStreetMap data with Python
30mCreate Detailed Traffic Models and Simulations
05mUsing OpenStreetMap with the Python OpenPoliceData package
01h 41mOpenStreetWhat? Mapping The World With Open Data