Downloaded 97 times
Uploaded byKnowledge Discovery and Data Mining
MAtlas: A case study on Milan mobility
The document analyzes GPS tracking data from 17,000 private vehicles in Milan, Italy over the course of one week. It finds that: 1) Traffic patterns followed consistent daily and weekly trends, with more traffic during weekday commuting hours and less on weekends. 2) Trip lengths and durations followed power law distributions, with many short trips and few long trips. 3) Traffic was concentrated in the city center during early morning and late evening hours, and dispersed to other areas during daytime commuting periods. 4) Clustering trips by route identified groups that took direct and indirect routes between origin and destination points like the city center and northwest areas.
More Related Content
Similar to MAtlas: A case study on Milan mobility
MAtlas: A case study on Milan mobility
- 1. MAtlas : acase study on Milano, Italy
- 2. Dataset info GPStraces 17K private cars one week of ordinary mobility 200K trips (trajectories) Milan, Italy Data donated by OCTO Telematics Italia
- 3. Overall view oftrips performed in a single day (Wednesday, April 4 th , 2007) Difficult to understand anything
- 4. Temporal analysis: intensityof traffic (n. of moving vehicles) per hour over the week The same double-peeked shape for all days, a bit lower in the weekends
- 5. Distribution of lengthsof the trips Neat power-law -> several short trips, few very long ones
- 6. Distribution of tripduration Another power-law, similar shape
- 7. How do lengthand speed of trips correlate? Average length grows with avg. speed (right plot) Yet, only slow trips reach considerable length (left)
- 8. Where is trafficconcentrated between midnight and 2 a.m.? (red = most intense)
- 9. Where is trafficconcentrated between 6 a.m. and 8 a.m.?
- 10. Where is trafficconcentrated between 6 p.m. and 8 p.m.?
- 11. Select only tripsthat start in the city centre (orange) and move to North-West Behaviours are still rather heterogeneous Notice the O/D matrix navigation tool on the right
- 12. Trajectory clustering dividestrips based on the route they cover Different color = different group Outliers are removed
- 13. Three sample clustersare highlighted One group (red) goes straight to NW, the others follow alternative routes
- 14. Temporal analysis oneach group tells us when they perform the trip A small group in the morning (commuters working outside the city?) a much larger one in the afternoon (incoming commuters?).
- 15. Origin/Destination analysis isflexible Analyze traffic from/to city areas to/from parking lots
- 16. Focus on aspecific (high frequency) parking lot, close to Linate airport
- 17. Analyze typical itinerariesfollowed to reach such parking lot T-Patterns -> overall view
- 18. T-Patterns: highlight onepattern that comes from the centre
- 19. T-Patterns: highlight onepattern that comes from North, along the “tangenziale” (ring road)
- 20. T-Patterns: highlight onepattern that comes from South, along the “tangenziale” (ring road)
- 21. Where is peoplebetween 6pm and 8pm of Wednesday, April 4 th ?
- 22. Where is peoplebetween 8pm and 10pm of Wednesday, April 4 th ? An high density spot appeared
- 23. Where is peoplebetween 10pm and midnight of Wednesday, April 4 th ? The dense spot disappeared. What happened?
- 24. Focus on thehigh-density spot Centered on the parking lots of the stadium April 4 th , 2007: a football match took place there...
- 25. Have a closelook at when people arrived to the stadium, and when they left Through O/D matrix tool, focus on traffic from/to stadium area
- 26. Arrivals and departuresdistributed as expected (concentrated resp. before and after the match) Small surprising result: some people start leaving around 30 minutes before the match ended...