Exploring Urban Mobility and Accessibility through Transport Data a study of London Oyster card data and disability
Gareth Simons, Dr. Andrew Hudson Smith, Dr. Martin Zaltz Austwick Centre for Advanced Spatial Analysis (CASA) University College London (UCL) London, UK
[email protected]
Stylianos Tsaparas School of Architecture University of Thessaly (UTh) Volos, Greece
[email protected]
Katerina Skroumpelou Computer Networks Laboratory, School of Electrical and Computer Engineering National Technical University of Athens Athens, Greece
[email protected]
Gianfranco Gliozzo Extreme Citizen Science (ExCiteS) research group University College London (UCL) and the Zoological Society of London (ZSL) London, UK
[email protected] I.
Abstract. This paper explores the accessibility of the London Underground network. To do so, we visualize and analyse TfL Oyster Card and Disabled Freedom Pass Oyster card data. We compare census data of people with limited mobility with accessible station usage. We explore travel patterns and network load during a typical week. We propose a new Android app that directs people with limited mobility to accessible stations nearby, to encourage their use. We explore different kinds of visualisation techniques with video and 3d animation. The visualisation approach to the analysis of these data proved very helpful in the attempt to understand the measure upon which public transport in London is used by people with limited mobility. The use of smart ticketing and data recording of the trips provided by TfL enable a thorough research of urban movement patterns, and allow various interpretations of
the city. If rhetorics towards a smart city are in place today, proclaiming more and more the need of smart technologies, sensors and a vision of a hybrid, cyber physical environment, the analysis of the data that this future urban state produces should be treated carefully. If used in the correct way, they will be able to reveal problems of the modern society that had remained in the dark, helping towards a vision of the desired urban well being for all. This paper can also serve as a portfolio of different takes on spatial data visualisation. Keywords: London Underground; Oyster Card; data visualisation; accessibility; spatial analysis; II.
INTRODUCTION
A. City and public space
The city is its people. It is the people it houses and the people it bears on its streets and infrastructure. People move within the city and its public spaces, venues of social interaction and economic exchange, which are the predominant activities that constitute cities. The city must be able to host and accept all of this movement and exchange within and around its public spaces, so that it can offer people the opportunity of being more active and socially engaged. The ever increasing size of cities also means that transportation infrastructure is increasingly essential for the mobility of citizens. It is, therefore, a fundamental role of the city to provide its residents with sufficient and equitable access to its streets, public spaces, and transportation. B. The issue of Accessibility If public space is not available for all, then it is no longer truly public. What defines a space as public is its accessibility. Many spaces are difficult to fully navigate, utilise, or feel comfortable in, unless citizens are young, ablebodied, and stereotypically “conventional”. The more a person deviates from this norm, the more inaccessible a space might be or seem. For our project, we chose to focus on people with limited physical mobility, such as wheelchair users. Observing London, we can see, seemingly everywhere, a significant attempt to make spaces, buildings, and public transport, accessible to all. Our goal is to explore to what extent these facilities are used in relation to the general population. It is an attempt to derive how effective the efforts towards improving accessibility have been. C. Public Transport for people with limited mobility In order to answer some of our questions, we explore the mobility of Disabled Freedom Pass Oyster card holders. This approach and the ensuing visualizations illustrate a number of perspectives and potential issues concerning the accessibility of the London Underground network. Our emphasis is on visualization as opposed to indepth technical analysis, and our findings are explorative rather than conclusive. We keep the process open for readers and viewers to arrive at their own conclusions. We know that our sample is limited, because not all residents of London that have a disability request an Oyster Freedom Pass. We also know that not all Disabled Freedom Pass holders use public transport. As mentioned in the website of London Councils, “There are approximately 1.4 million disabled people in London, though precise figures are unknown” and “There are over 1.3 million Freedom Pass holders (of which 1.16million are older people and 160,000 are disabled). Those who are of the eligible age or meet the criteria for a disabled pass and whose sole or principal residence is in London are entitled to the pass”. In other words, only 8.75% of Londoners with disabilities are Freedom Pass holders. Our work is therefore based on
how accessible London’s public transport network is for the people who intend to use it. D. Data We obtained a 5% sample of Oyster Card trip data for a week in November 2009. The 5% sample, which translates into 2.5 million trips, is still sufficient for gaining a substantial view of Oyster card usage patterns. The data contains the following information: Day of the week; Subsystem used (bus, underground, tram, overground, dlr, national rail, trips to Heathrow); Start and ending stations (except for bus trips); Entry time and end time of journey; Bus route; And the type of oyster card user (travelcard, pay as you go, staff pass, freedom pass elderly and disabled, bus and tram pass). III.
THE ACTIVE “CITY” BEYOND BARRIERS.
E. Commuters and citizens. The distribution and ratio of disabled to the non disabled is geographically visualised to explore the dynamic spatial and temporal patterns that emerge. People are constantly moving from place to place and thereby changing the composition of population in space and time in proximity to stations. We here establish a link between aboveground and below ground data by linking census data with TFL data. F. Disability in census 2011 TFL offers two versions of Freedom Passes, one being for elderly, and the other for the disabled. For our exploratory analysis, we assume an overlap between those that are eligible for the Disabled Freedom Pass, and those that report in the census their “Daytoday activities limited a lot” or “daytoday activities limited a little”. These categorisations require a longterm health problem or disability of duration greater than 12 months. These categories therefore do not directly include elderly people unless they have a specific disability or significant agerelated ailment of significant duration. The fact that Freedom Passes granted for old age are easier to obtain than those given to people with a disability, will create a possible misalignment between the two datasets. A further distinction between the two datasets is that the TFL data includes trips made by nonresidents1. In general terms, our exploration assumes that: Assumption 1: (Daytoday activities limited a lot + Daytoday activities limited a little) = Eligible for freedom pass. We investigate at the MSOA2 level the spatial relationship amongst populations with different percentages of people reporting disabilities. Some patterns are visible, where the more central areas that It influences only the total; number of travellers since non residence cannot get any freedom pass 2 Middle layer super output areas (MSOAs) 1
have better transportation access are characterized by a lower degree of population reporting limitations in their daytoday activities. G. The construction of the geodata The analysis proceeds to create catchment areas around all tube stations to explore workflows and gain insight from the merging of the two datasets for investigating opportunities for visualisation. A buffer of 2 Km around stations identifies an approximate area served by all stations (Figure 1). Fig. 1. Stations (red dots) and buffered area
Fig. 2. Initial status (0am Saturday). Visualizing only census data
Then: Assumption 2: Every station has an exclusive “catchment area” The space around each of the stations is assigned to each station by using a Voronoi tessellation. The Voronoi and buffer are then linked with census data by using centre points for the Output Areas, which are the smallest spatial units available in the census. For every catchment area, the census data for the resident population and residents with longterm health problems or disabilities limiting their activities by alot or alittle, are summarised. This is used as a starting point for exploring how the resident populations may vary throughout the course of the day based on tube trip data. The dynamics of these daily population movements is explored using Processing, into which the data and geospatial information are imported and animated. The colorcoded areas change during the day, mainly in central London, indicating indeed a change in the dynamic population (Figures 2 and 3).
Fig. 3. After some hours, the situation changes, but mainly in the central areas (situation after 31 hours, 7am Monday). IV. TRAVEL PATTERNS / REVEALED
H. Exploring the background This Processing application has the purpose of revealing travel patterns for Disabled Freedom Pass holders (DFPH) and comparing them to nondisabled Oyster card users (NDOCU). When launched, on the left half of the screen, the application shows trips as lines from a station to another throughout each minute of each day of a week. The trips have different coloring, with red representing DFPH trips and white representing DFPH (Figure 4).
Fig. 4. Screenshot of the app
On the right half of the screen, seven graphs are plotted, representing the days of the week from Sunday to Saturday. The length of the vertical lines of the graph represents the load of the tube at each time step. Again, the white lines represent NDOCU and the red DFPH. As the week progresses, we clearly see that the white lines have notable peaks during the morning and afternoon rush hours, whereas the red lines have no peaks. This is likely because people with limited mobility may have a tendency to avoid use of the Underground if other options are available to them, particularly during peak times, and as indicated in the data, 83.9% of DFPH trips are made on London Transport Buses. Rush hours can be stressful and time consuming for people with disabilities. J.D Schmockera et. al., 2008, found that people with limited mobility “value their comfort more, due to reduced ability to move easily” and even if they are offered a freedom pass “there appears to be a preference for modes that offer more independent mobility”. They also note that “While rail and underground (tube) services are widely available, most are not readily accessible for wheelchair or electric scooter users.”
route code. The resultant data is approximately 750,000 individual trips. After all bus lines were removed, we combined the new dataset with a dataset containing the locations of station positions. This was done using the MS Access query wizard to assign geographic station locations to each trip, resulting in a dataset containing four extra columns; two for Easting and Northing position of the starting station; and two for Easting and Northing position of each ending station. The next step was to calculate the load of DFPH and NDOCU for each station, and also calculate the means and percentages. To do these calculations we used R (Figure 6).
Fig. 6. Disabled Freedom Pass Holders’ load on tube stations (the radius of each circle represents the percentage) V.
Fig. 5. The load of passengers leaving Stratford station, compared to the mean load (the black line)
The app is interactive, and by hovering the mouse over each station, the user can see the name of the station, whether the station is accessible or not, and two linegraphs (Figure 5). One shows the load of the station compared to the mean load of all stations (thus, its popularity) for DFPH and NDOCU. These graphs help us understand how much a station is actually used, also compared to its general load of passengers. Finally, another two options are added, where the user can choose to show all stepfree access stations, or all stations ranked according to their total load of passengers (where the radius of each circle represents the percentage). I. Methodology We needed to add spatial information to the data to plot it on a map. So, the first step was to remove all rows that did not have a specific geographic reference, which includes all data rows referring to bus trips, since the only information contained in these was the bus
ACCESSLONDON – AN ANDROID APP
J. Concept Based on initial visualisations, we observed that many accessible tube stations were not used as much as we had anticipated. We therefore thought that it may be beneficial to create an app pointing the user towards the closest accessible stations along with pertinent information about the station and context. The idea continued to evolve and nearest bus stop information was also added, based on buses’ popularity and also due to the fact that all buses offer disabled access. As found by Kim and Ulfarsson (2004), “the distance between residence and nearest bus stop influences the mode share” and since walking time to nearest public transport facilities is highly valued (Iseki et al., 2012), given the ease of guidance to the nearest station in the form of a mobile app, may make transportation choices easier. There are many applications available that focus on trip planning. However, this one is different in the sense that it is focused on finding the nearest accessible stations and bus stops for the disabled users. The process is made fast and simple by requiring just two clicks (taps).
Fig. 7. Screenshot, home page of the app
To add to this concept, the user also has live updates from the twitter accounts of @TfLAccess and @FreedomPassLDN, two channels that deal with accessibility of public transport in London, on which users can make queries that are answered on a frequent basis. As an extra, informative feature, the app gives users data on the popularity of each mode of transport. The app is currently focused on wheelchair users, but an idea for future development is to incorporate different application modes and information types to meet the requirements of different disabled users. K. Methodology The app uses three static data sets. The one contains the locations of accessible tube stations. The other contains the locations of bus stops. The third contains information about each mode of transport. This information is the percentage of users that the specific mode carries, and the percentage of DFPH this mode carries. It was a challenge getting the app to work because of idiosyncrasies of the android mode of Processing. It was further challenging connecting to the twitter API, and finding the simplest way of getting the app to automatically load the map application of the phone. VI. IDENTIFIED FLYING OBJECTS
L. Background The question behind the third part of our portfolio is: “Is there an effective way to visualise the relationship between the (underground) spatial movement of people with disabilities and the (above ground) built environment?”. In this case, the message is the information that our data encapsulates, and which is revealed through an exploration of key points that influence the nature of our visualisation methodologies.
The challenge is therefore twofold: understanding and revealing the nature of the data; and developing visualization methodologies that are the most capable of releasing this information. To begin with, the nature of our data implies the temporal duration and spatial distance of each trip from its starting to ending points. These points are located within the fabric of the urban environment, but the movement between these points occurs in a void of spatial context. We are therefore seeking a way to visualise this movement in a manner that permits a greater comprehension of temporal duration and spatial distance, and which can therefore be perceived in a more engaging way. There are many ways to depict the city. The potential variances between what is depicted and what the observer perceives is narrowed in the case of a picture, or even better, a video of the city. The concept thus lends itself to representing the underground movements of people aboveground instead, where the animated patterns of movement contrast with the stationary built environment, while simultaneously making the connection between underground and aboveground locations.
Fig. 8. Kilometers travelled by the two groups of passengers
M. Description At the bottom of the display there is a line with three attributes: the time, the mean straightline distance that people with disabled Freedom Pass cardusers travel per day, and the mean straightline distance that other card holders travel per day. The interesting result is that general trips move in a radius of moreorless eight kilometres whereas disabled Freedom Pass cardusers move an average of seven kilometres per day (Figure 8). This observation can lead to several hypotheses, however we will refrain from making overly general or speculative assumptions. N. Methodology The project Identified Flying Objects was built using Blender v.2.70, Adobe Premiere video editor and the Visual effects (VFX) processes (Zwerman, Okun, 2010). VFX are display techniques which combine live action scenes with generated imagery in order for a realistic environment to be composed (Brinkmann, 2008). The choice of the VFX method for the visualisation of the current project was based on three factors: First of all, as it is referred above, the wish to convey the nature of the data in an engaging and familiar contextual environment. Secondly, the opportunity to explore novel visualisation techniques that are not used widely in scientific analysis. Finally,
the choice to use reallife video rather than building a detailed digital 3D model of London allowed for a realistic outcome and efficient workflow. The VFX process consists of three welldivided sections. At the beginning of the first section is the video recording, which was performed with an iPhone 5. Blender can only recognize two movement planes for a camera’s path. In no cases can it handle a combination of movement in 3D space and rotation around a point. The next step is to setup the digital camera in the Blender environment. This particular camera recognizes the path in the recorded video by tracking random points across the whole video frames. In this way, the digital camera performs exactly the same motion as the regular camera, thus accurately depicting the digital rendered output. Importantly, the digital camera must be set up with the same technical specifications of the video camera, including factors such as focal length, angle of view, lens distortion, etc (Figure 9).
the video of London. Due to the setup in step 1, the start and end points of the flying objects are correlated to the video. The third part of the VFX process is the modification of the final output in terms of a realistic and illustrated visualisation. This includes lighting and texture settings. A significant role was played by Adobe Premiere in generating the final edit of the video, and was used for visual effects such as blur, adding sound, and for reducing the length of the clip to fit the needs of the presentation.
Fig. 10. The final result VII. CONCLUSIONS AND THE USE OF UNITY
Fig. 9. Adjusting the factors
The second step is the combination of the video with the information derived from the dataset. The data was imported into blender using code scripted for the blender python “bpy” API. The script consists of the following steps: Importing all data and creating trip data “objects”, which are stored in a trip “dictionary”; Creating a default trip object blender material; Setting up a starting point for the blender scene, including a base plane and lighting; Optional (not used for the final rendition) smoke trails; Iterating through the tripobject dictionary to create an object for each trip; Whilst iterating, each trip object is keyframed to invisible, then visible at the time the trip commences, then starting point, then ending point, and then invisible again. Note that because the objects are not instanced and destroyed “onthefly” there are some practical limitations to the number of objects that can be imported. The plane and the objects constitute the 3D digital environment of our model which is overlaid with
O. The insandouts of different visualisation software. Each of the visualisation strategies employed so far reveals a different natural fit for the exploration, visualization, and interaction with data. Unity occupies a unique position because it offers a degree of ‘real time’ interaction and performance not matched by the other approaches. As a game engine, it is designed from the groundup for this purpose, thereby offering a unique range of benefits: 1) It offers a modular approach towards “assets” and “resources”, allowing for the flexible arrangement and combination of data, 3d models, settings, and scripts; 2) It separates ‘real time’ from the frame rate, which means that the frame rate is constantly optimized for a device’s computational performance, without leading to wildly fluctuating timestep changes; 3) It is capable of handling a serious quantity of objects in realtime. Testing with minimal rendering requirements indicated the ability to manipulate well upwards of 3,500 objects depending on computational power and the time scale;
4) It is further capable of offering suffiently high quality graphics rendered in realtime, therefore distinguishing itself from traditional rendering and animation engines which can be notoriously slow at rendering, albeit with increased realism. 5) Due to these and other reasons, it is inherently wellsuited to the creation of dynamic and interactive visualisations that actively respond to user inputs. P. Data Prep The data preparation for Unity was done in Python and took three inputs: 1) The tube lines with each of the stations; 2) The stations names with the station coordinates; 3) The data file consisting of all trips. The script creates a station index, and a station coordinates list, which it then uses to create a weighted adjacency matrix of all stations on the network. The scipy.csgraph package is then used to return a solved shortest path array. Subsequently, the data file is imported and the start and ending location for each trip is then resolved against the shortest path array, with the resulting waypoints for each trip written to a new CSV file. Q. Methodology The Unity implementation consists of various components: 1) The 3d models for the London landmarks were found in the Sketchup 3D warehouse. Their materials were removed and they were imported to Unity in FBX format; 2) The outline for Greater London was prepared as a shapefile, which he subsequently exported to FBX via City Engine; 3) An empty game object is assigned with the main “Controller” script that provides a springboard to other scripts and regulates the timescale and object instancing throughout the visualisation. This script allows numerous variables to be set via the inspector panel, including the maximum and minimum time scales, the maximum number of nondisabled trip objects permitted at one time (to allow performance finetuning), a dynamic time scaling parameter, and the assignment of object prefabs for the default disabled and nondisabled trip objects. Further options include a moviemode with preset camera paths and a demo of station selections; 4) One of the challenges in the creation of the visualisation was the need to develop a method for handling time scaling dynamically to reduce
computational bottlenecks during rushhours, and also to speed up the visualisation for the hours between midnight and morning to reduce the duration of periods of low activity. The Controller script is therefore written to dynamically manage the time scale; 5) The controller script relies on the “FileReader” script to load the CSV files. The stations CSV file is used to instance new station symbols at setup time, each of which, in turn, contain a “LondonTransport” script file, with the purpose of spinning the station symbols. It also sets up behavior so that when a station is clicked, the station name is instanced (“stationText” script) above the station, and trips only to and from that station are displayed via the Controller script. The FileReader script also reads the main trip data CSV file, and loads all trips at setup time into a dictionary of trip data objects that include the starting and ending stations, as well as the waypoint path generated by the Python script. The trips data objects are then sorted into a “minute” dictionary that keeps track of which trips are instanced at each point in time. The minute dictionary is in turn used by the Controller script for instancing trip objects. 6) The “Passenger” and “SelectedPassenger” objects and accompanying script files are responsible for governing the appearance and behavior of each trip instance. Since thousands of these scripts can be active at any one point in time, they are kept as simple as possible, effectively containing only information for setting up the trip interpolation based on Bob Berkebile’s free and open source iTween for Unity3. iTween is equipped with easing and spline path parameters, thereby simplify the amount of complexity required for advanced interpolation. The trip instance scripts will further destroy the object once it arrives at the destination. 7) Other script files are responsible for managing the cameras, camera navigation settings, motion paths for the movie mode camera, rotation of the London Eye model, and for setting up the GUI. R. Visual and Interaction Design It was decided to keep the London context minimal with only selected iconic landmarks included for the purpose of providing orientation, and a daynight lighting cycle to give a sense of time (Figure 11). Disabled Freedom Pass journeys consist of a prefab 3
http://itween.pixelplacement.com
object with a noticeable bright orange trail and particle emitter, in contrast to other trips which consist of simple prefab objects with a thin white trail renderer and no unnecessarily complex shaders or shadows due to the large quantities of these objects. The trip objects are randomly spaced across four different heights, giving a more accurate depiction of the busyness of a route, as well as a more threedimensional representation of the flows. Interactivity is encouraged through the use of keyboard navigation controls for the cameras, as well as a mouse “look around” functionality, switchable cameras, and the ability to take screenshots. When individual stations are clicked, then only trips to or from that station are displayed.
choice of older and disabled people: a case study of shopping trips in London”. Journal of Transport Geography, Volume 16, Issue 4, July 2008, Pages 257–267 [6]
Kim, S. and Ulfarsson, G.F. (2004). The Travel Mode Choice of the Elderly: Effects ofPersonal, Household, Neighborhood and Trip Characteristics. Paper presented at the 2004 Annual Meeting of the Transportation Research Board, Washington,
[7]
DC.Hiroyuki Iseki, Brian D. Taylor, Mark Miller. The Effects of OutofVehicle Time on Travel Behavior: Implications for Transit Transfers, Tool Development to Evaluate the Performance of Intermodal Connectivity (EPIC) to Improve Public Transportation. California Department of Transportation, Division of Research and Innovation, 2012
[8]
Ketai Library for Processing (https://code.google.com/p/ketai/)
[9]
Twitter4j A java library for the twitter API (http://twitter4j.org/en/index.html)
Android
[10] Brinkmann (2008) The Art and Science of Digital
Compositing: Techniques for Visual Effects, Animation and Motion Graphics. 2 edition. Amsterdam ; Boston, Morgan Kaufmann. [11] Zwerman, S. & Okun, J.A. (2010) The VES
Handbook of Visual Effects: Industry Standard VFX Practices and Procedures. 1 edition. Burlington, MA, Focal Press. [12] http://itween.pixelplacement.com
Fig. 11. Screenshots of the app REFERENCES [1]
http://www.londoncouncils.gov.uk/londonfacts/defa ult.htm?category=2 [2]
http://www.londoncouncils.gov.uk/services/freedom pass/faqs/ [3]
geoMap library (http://www.gicentre.net/geomap/) from the giCentre, City University London.
[4]
Image with accessible http://www.waag.co.uk/awards.asp
[5]
JanDirk Schmöckera, Mohammed A. Quddusb, Robert B. Nolanda, Michael G.H. Bella. “Mode
logo
from