This document provides an overview of bike share programs around the world and efforts to visualize their data. It discusses various bike share systems, challenges with data collection, and analyses of trends like peak usage and bike/dock ratios in different cities. The author describes their website visualizing real-time bike share station and availability data for several cities. Community efforts to analyze and build APIs and apps for bike share data are also mentioned.
Dubai Call Girls O528786472 Call Girls Dubai Big Juicy
Visualising Bike Share Systems Around the World
1. Visualising Bike Share
From Boris Bikes in
London to Bike Share
across the World
Oliver O’Brien
UCL CENTRE FOR ADVANCED SPATIAL ANALYSIS
Photo CC-NC-By-SA Adam Bowie on Flickr
2. Contents
• Intro to bike share
• Around the world
• Visualising
• Analysing trends
• Other community
efforts
Photo CC-By Charlotte Gilhooly on Flickr
3. What is a
Bike Share?
• A scheme allowing
bikes to be hired from
(and returned to)
certain locations
• City or campus based
• Typical use is for short
durations
• Generally fully
automated (in theory)
• Require an account
linked to a credit card
Photo CC-NC-By-ND Terry Freedman on Flickr
4. • Docks – The things
which hold onto the
bikes and release them
• Stations – groups of docks
• Spaces – docks which are empty
Photo CC-By Les Hutchins on Flickr
Terminology
5. The Bikes
• Normally “odd”
looking, “lively” colours
– Eye catching amongst
street furniture
– For sponsor branding
– To discourage theft
• Custom designed with
non-standard parts, to
prevent part theft.
Photos CC-By tsuacctnt and CC-NC-By-ND Monica Vidal on Flickr
7. Current Locations for my Visualisation
City Official Name Installed System # of Bikes
London Barclays Cycle Hire July 2010 Bixi 4,300
Barcelona Bicing March 2007 Bikemi 4,200
Milan Bikemi December 2008 Bicing 1,100
Saragossa Bizi May 2008 Bicing 800
Girona Girocleta September 2009 TNT 100
Washington DC
and Arlington
Capital Bikeshare September 2010 Bixi 650
Montreal Bixi May 2009 Bixi 4,200
Minneapolis Nice Ride June 2010 Bixi 600
Denver B-cycle April 2010 B-cycle 350
Melbourne Bike Share June 2010 Bixi 400
8. Rejected Locations
City Official Name System # of Bikes Reason
Shanghai Forever Forever 50,000 Rate limiting
Stockholm City Bikes Bicing 1,000 Blocking the data
Cardiff OYBike OYBike 100 Key changes at midnight
Lyon Velo’V Velib 3,000 Scraping proving difficult
Paris Velib Velib 17,300 Take-down request
Brussels Villo Velib 1,700 Take-down request
Dublin dublinbikes Velib 400 Take-down request
Valencia Valenbisi Velib 1,000 Take-down request
Seville Sevici Velib 1,850 Take-down request
Vienna Citybike Velib 750 Take-down request
Toyama Cyclocity Velib 130 Take-down request
Brisbane CityCycle Velib 500 Would result in take-down request
9. To Be Added
City Official Name Installed System # of Bikes
Mexico City Ecobici February 2010 Bicing 1,000
Rio Samba November 2009 Samba 100
Torino Tobike June 2010 Tobike 1,200
Dijon Velodi June 2008 Bikemi 350
• Any others which have websites with location data
for stations and counts for both bikes and spaces
– Not Nextbike schemes in Eastern Europe (no spaces)
• But they do have bike IDs for up to 5 bikes at each dock
– Not various large schemes in China (no website)
– Not Velib (as requested by operator)
10. Melbourne – Helmets
• Scheme started in June
2010, slow to grow
– By law, helmets must be
worn (or AU$150 fine)
– Helmets are not supplied
with the scheme
– Can now buy AU$5 helmet
from two vending machines
or a supermarket chain
• Return to a supermarket for
AU$3 cash-back
– Also launched in the middle
of winter
Source and image from bike-sharing.blogspot.com
11. Brussels – Where’s My Villo?
• Campaign group
– Aiming to improve:
• Reliability
• Distribution
• Service level transparency
– Tracking performance measures
– Interested in comparing with
other cities
Source and images from wheremyvillo.be
12. Denver – Conspiracy
• “Republican gubernatorial candidate Dan Maes is warning
voters that Denver Mayor John Hickenlooper's policies,
particularly his efforts to boost bike riding, are ‘converting
Denver into a United Nations community.’
“Dan Maes said Denver's
B-Cycle bike-sharing
program was promoted by
a group that puts the
environment above citizen
rights.”
– article in the Denver Post
Source denverpost.com, photo CC-By-NC Trace Altman on Flickr
13. Let’s Visualise Them!
• Obtain the data from the operators’ websites
– Some provide XML/JSON/KML
– Lots of Regex parsing
– Velib-based systems require two stages
• Store it for analysis
• Stick it on a map
– OpenLayers has some nice vector styling for points
– OpenStreetMap-based background
– Charts of historical trends via the Google Chart API
'id:"([0-9]+?)".*?name:"(.+?)".*?lat:"(.+?)".*?long:"(.+?)".*?nbBikes:"([0-9]+?)".*?
nbEmptyDocks:"([0-9]+?)".*?installed:"(.+?)".*?locked:"(.+?)".*?temporary:"(.+?)".*?'
16. Animation over 48 hours
• All in Javascript
– Using SVG (or VML in Internet Explorer)
• The animation is extremely slow in I.E.
• Not great in Firefox
• Excellent in Chrome/Safari
http://oobrien.com/vis/bikes/timeline.php?city=london
17. N Europe:
1. London
2. Paris
3. Dublin
4. Brussels
Spain:
1. Barcelona
2. Girona
3. Valencia
4. Seville
America:
1. Wash. DC
2. Montreal
3. Minneapolis
4. Denver
1. Vienna
2. Milan
3. Toyama, JP
4. Melbourne
18. Bike/Dock Ratio
• A key component in the optimisation of a bike hire
scheme
• For the users, having too many bikes is very bad
– Frustrating if you can’t drop off your bike while the clock
is ticking.
• But more bikes mean more visibility for the
scheme and promotion for the sponsors
19. Bike/Dock Ratio
• No of bikes per 100 docks
– Based on max availability
at around 5am (“no” usage)
– Averaged over a few weeks
City Ratio/100
Melbourne 60
London 56
Montreal 56
Denver 54
Milan 52
Dublin 51
Minneapolis 50
Toyama 50
Barcelona 49
Washington DC 49
Girona 48
Paris 47
Vienna 47
Brussels 46
Seville 42
Valencia 39
Average 50
Background map CC-By-SA OpenStreetMap contributors Preliminary/unreviewed data
20. Peak Usage % (Weekday)
• Max % of bikes being used
– Data from last Wednesday
– Not directly measurable
– Assumes that usage
dropped to zero overnight
– Simple analysis, not
considering the effect of
weather conditions, public
holidays or special events
City Peak
Dublin 41%
London 25%
Valencia 22%
Girona 21%
Barcelona 20%
Seville 20%
Milan 18%
Paris 15%
Montreal 13%
Melbourne 12%
Washington DC 11%
Brussels 10%
Toyama 9%
Vienna 9%
Denver 8%
Minneapolis 6%
Photos CC-NC-By-ND D1v1d on Flickr Preliminary/unreviewed data
21. Peak Usage % (Weekend)
• Max % of bikes being used
– Data from last Saturday
– Weekend usage much
higher than weekday usage
for the U.S. cities, lower for
Europe
City Peak
Dublin 25%
Barcelona 20%
Washington DC 19%
Denver 18%
Girona 17%
Valencia 16%
Vienna 12%
Seville 11%
Milan 11%
Minneapolis 11%
London 10%
Paris 10%
Montreal 8%
Brussels 7%
Melbourne 5%
Toyama 3%
Photos CC-NC-By DDOT DC on Flickr Preliminary/unreviewed data
22. Bike-o-Meter
casa.ucl.ac.uk/bom
• Tweet-o-Meter for bikes
– Steven Gray (@frogo)
– Using Google Gauges
• See the real life Tweet-
o-Meters at the new
British Library “Growing
Knowledge” exhibition
– Should be easy to hack
to show the Bike-o-
Meters instead
23. Weekday Use – 1. Europe ex-Spain
Preliminary/unreviewed data
28. Weekend Use – 3. Rest of World
Preliminary/unreviewed data
29. More Analysis
• London
• Graph shows
number of bikes
available to hire
• Effect of rain
– Using the CASA
weather station
• Effect of the
tube strikes
Preliminary/unreviewed data
30. More Analysis!
Clustering
• Geodemographics of
a city area based on
usage patterns of
stations within it?
• Could combine with
existing demographic
data to predict likely
usage patterns of
new stations
Clustering output courtesy of James CheshirePreliminary/unreviewed data
31. Even More Analysis!
Redistribution Effectiveness
• Distribution
– Which cities have the most effective redistributions?
– When does the redistribution occur?
– Does it actually make things worse?
33. Even More Analysis Possible?
• Shapes and sizes of cities and their schemes
– How convenient is the scheme for the intended users?
– Coverage in residential versus commercial areas
• Dock station densities
– How far away from your destination to you need to go to
find a docking station?
• Flows
– Would require “bike-level” information rather than
“station-level” as at present
34. Also in the Community
• Adrian Short (@adrianshort)
– first “Boris Bikes” API for London
– http://borisapi.heroku.com/
• Andrew Larcombe (@andrewl)
– Where Are The Bikes API - A universal PHP API for
extracting data for bike share schemes
– Currently includes over 60 schemes
– http://github.com/andrewl/watb/
35. Also in the Community
• Aidan Slingsby (City Uni)
– www.gicentre.org/tfl_bikes/
• Includes “seeing ahead” for
the next four hours
• Tom Taylor
– Cycle Hire Explorer
• Includes total usage counts
• cyclehire.tomtaylor.co.uk/
• Lots of cycle hire apps
for iPhone/Android
Screenshot of Aidan Slingsby’s TFL Bikes graphs
36. Also in the Community
Screenshots of some of
the apps on the iPhone
(iOS4) for the London
Cycle Hire scheme.
Clockwise from top left:
Cycle Hire Live,
iLondonCycle, London
Cycle, Bixou Lite, Blue
Lanes, Apple App Store
37. Thanks!
Email: o.obrien [at] ucl.ac.uk
Blog: oliverobrien.co.uk
Twitter: @oobr
Photo CC-NC-By Kurtis Garbutt on Flickr
Notas del editor
The picture is of the T-shirt that was given out to the first 1000 people that signed up for the scheme in London.
Fully automated, including sign-up and operation. However London, at least, employs workers at some of the stations at rush-hour to provide extra capacity (bikes in the morning, spaces in the evening). This is a short term measure. Waterloo will shortly get 350 docks which should alleviate this problem.
This is a familiar scene to commuters leaving the terminal stations in the morning, or leaving the City in the evening...
Above: Minneapolis “Nice Ride”. Below: Barcelona “Bicing”
The cities are at roughly the same scale (exactly the same zoom level, which = scale * cosine (latitude))
Screenshot of empty bike stations in west Barcelona at 9pm on Wednesday.
Note this is just one day’s worth of data – and doesn’t control for weather conditions or special events.
Note this is just one day’s worth of data – and doesn’t control for weather conditions or special events.
London, Dublin and Brussels have a slightly earlier evening peak than Milan and Paris. Vienna doesn’t use the scheme to commute.
Use at lunch is as much as (or more than) during the evening rush hour, and the evening peak occurs at around 7pm – the Spanish “siesta” working pattern”. Girona is very small, so usage rises gradually during the day, rather than having a morning rush hour.
Denver doesn’t have the rush hour peaks – showing that few commuters are using the scheme, possible due to the U.S car culture?
Usage in general much lower than during the week, and mainly late afternoon use, particularly in Paris and Milan.
Spain still has a siesta at the weekends! (The effect is noticeable in four of the five Spanish cities included.) The bikes also get used all through the evenings.
Higher usage at the weekend than during weekdays, in the US.
Comparing world city dynamics based on bike shares!