Future Cities Hackathon 2013

Via DataScienceLondon, we won 'honourable mention' for our proposed dynamic pricing model of on-street parking bays in central London: http://ianozsvald.com/2013/10/07/future-cities-hackathon-ds_ldn-oct-2013-on-parking-usage-inefficiencies/

Transcript

1. www.morconsulting.c Future Cities Hackathon 2013 (DataScienceLondon) Modelling parking bay utilisation and proposing an adaptive pricing model Amit, Bart, Dominic, Ian, Jackie, Mateusz
2. FutureCitiesHackathon 2013 (DataScienceLondon) Goal • We model parking bay usage • Are bays used efficiently? • Could people park more intelligently? • Could we price adaptively to encourage a change in behaviour?
3. FutureCitiesHackathon 2013 (DataScienceLondon) About Us • Amit Nandi • Bart Baddeley (littlebigdata.co.uk) • Dominic Steinitz (idontgetoutmuch.wordpress.com) • Jackie Steinitz • IanOzsvald.com (@ianozsvald) • Mateusz Łapsa-Malawski (chayamuni.com)
4. FutureCitiesHackathon 2013 (DataScienceLondon)
5. FutureCitiesHackathon 2013 (DataScienceLondon) Outline of our method • Take 1 day of data • Extract lat,lng pairs (>=1 per street) • Count start and end usage events • Model change by hour • Note over-usage (council gets free cash!)
6. FutureCitiesHackathon 2013 (DataScienceLondon) “Parking start” events on 1 day • Note 9am, 1pm, lunch lull
7. FutureCitiesHackathon 2013 (DataScienceLondon) Model of arrivals and expiries • 9am peak, lunch, evening, overcapacity
8. FutureCitiesHackathon 2013 (DataScienceLondon) Fines by day • Monday-Sunday • Penalties by hour
9. FutureCitiesHackathon 2013 (DataScienceLondon) Live occupancy data is coming • Live data will increase visibility of available parking spaces • Smartphone App “ParkRight” in App Stores • Small data sample available today for analysis
10. FutureCitiesHackathon 2013 (DataScienceLondon) What do we know? • We know which bays are heavily used or lightly used from Cashless Meters • We know when fines get handed out (and where if geocoding were added) • We have a pretty good idea about uneven resource usage • Where do the parking bays cluster?
11. FutureCitiesHackathon 2013 (DataScienceLondon) Location of bays in Westminster • Capacity/Spaces 1...36 (light to dark)
12. FutureCitiesHackathon 2013 (DataScienceLondon) Zoomed to Trafalgar Square • Amit Nandi •
13. FutureCitiesHackathon 2013 (DataScienceLondon) Smartphone app for parking? • Notify users of nearby (probably) underused spaces • Slice by time of day, warn of the number of fines issued nearby at this time of day • Integrate paid local advertising with useful Council messages in the App? • Next diagram white == low occupancy
14. FutureCitiesHackathon 2013 (DataScienceLondon) Underutilised parking spaces
15. FutureCitiesHackathon 2013 (DataScienceLondon) Confusing pricing for the bays!
16. FutureCitiesHackathon 2013 (DataScienceLondon) Proposal for adaptive pricing • Make under-utilised parking cheaper and easier to find • Encourage drivers to park in less-used streets (it relieves congestion) • Encourage Traffic Management Agents to target trouble areas (increases income)
17. FutureCitiesHackathon 2013 (DataScienceLondon) Thank you! • The code is all online: https://github.com/idontgetoutmuch/ ParkingWestminster (Apache license) • This presentation and images are all Apache licensed