As a cyclist in London you can do your best to avoid left turning buses and dozy pedestrians. One thing you can’t really avoid though is pollution (although I accept cyclists probably aren’t much worse off than pedestrians and drivers in this respect). To illustrate this I have taken data for 3.2 million journeys from the Barclays Cycle Hire scheme and combined it with GLA pollution data for particulate matter. Unsurprisingly, pollution is worse at junctions and where there is lots of static traffic, with the popular cycling routes around Waterloo Bridge and the Strand particularly affected. Most of the journeys are subject to relatively low (by London standards) levels because cyclists try and avoid the busiest routes, such as Euston Road. The loop around Hyde Park is really popular with Boris Bikers and fortunately one of the least polluted but clearly more could be done to sort out the pollution hotspots around the west end.

The routes have been guessed using routing algorithms and OpenStreetMap data and optimised for cyclists (ie we assumed that people would prefer cycle lanes over roads etc). Thanks to Ollie O’Brien for this analysis. You can see more of his work here. I produced this map using the R software package and blog about how I did it here.

Transport for London have just released their performance data (link here) for the London Underground network. It is in the form of a really detailed file that contains, amongst other things, the “Peak Operated Kilometres” and “Peak Passenger Journeys” for the past 6 years or so. If you total the distances covered by the Tube rolling stock at peak times, this map shows how far they travel each day. The Central Line wins with over 13,000km – the equivalent of almost reaching Australia! In humble last place is the Waterloo and City Line that just passes Dublin with a little over 500km travelled. There are many more cool things to be done with the data, but I thought this is a neat way of showing the gargantuan task of shifting London commuters!

This week I am giving a talk on some of the London maps that we produce in CASA. The hours of work I put in to such maps is minuscule compared to the amount of effort and time that the OpenStreetMap community have invested in producing a truly open (and often more accurate) map of London (and the rest of the world). To give an idea of just how far the project has come over the past five years or so, ITO World have produced a great visualisation (above).

As a bonus ITO World have also produced a similar visualisation for Berlin, Germany.

Thanks to Patrick Webber for the tip off about these.

*This post has been cross-posted from the Mapping London blog.*

A few months back I had the honour of being asked to approve the use of a couple of excerpts from my London Surname Map in The Times Atlas of London. The wait was finally over last week when I received my copy in the post. It is a great book and an essential guide to the city. The Atlas begins with the first mapped representation of London in 1250 and covers a detailed history of London mapping that provides interesting context to both the development of the city and its mapping culture.

The Atlas includes detailed information on each of the London boroughs interspersed with great photos and fascinating insights into London life covering everything from its inhabitants (below) to football,

and public transport (below) through to London’s World Heritage Sites.

As you would expect from an atlas there are may detailed maps (see below) to break up the extensive commentary associated with the topics I mentioned above. I think the Atlas is brilliant book (not just because I’m in it!) and should make a great addition to the coffee table of fellow London map addicts.

]]> http://spatialanalysis.co.uk/2011/10/the-times-atlas-of-london/feed/ 0 http://spatialanalysis.co.uk/2011/08/mapping-gcse-scores/ http://spatialanalysis.co.uk/2011/08/mapping-gcse-scores/#comments Tue, 16 Aug 2011 15:34:59 +0000 James http://spatialanalysis.co.uk/?p=2815 In the UK, August is exam results month for 16-18 year olds. Every year, photos of leaping teenagers clutching their results are accompanied by reports of record attainment rates, debates around how challenging modern exams are and, more so recently than ever, concerns for the number of sixth form and university places. Back in March the full list of the 2010 GCSE results (exams taken by UK 16 year olds [except in Scotland]) were released and I mapped them but never got round to sharing them with anyone. Now seems a good time to do this so here goes…

The map below uses the increasingly popular cartogram method to show the success of students in each Local Authority (LA) across England. The non cartogram version is also shown alongside.

This is quite a coarse map as England is only split into the 152 LAs and we know there is much greater variation between schools at a local level and even sometimes within individual schools. Moreover, schools on authority borders often serve communities from the areas on either side, limiting the application of LA data to their populations only. Independent (fee-charging) schools are also included in these broad LA results, which is significant when we take into account the predictably higher results of fee-paying pupils and the fact that these schools have not been established with regard for even distribution across the country.  The size of the LA (in school-age population terms) does not seem to have a strong link to the results of its pupils. There must be other factors at play. Concerning known evidence indicates that a pupil’s level of deprivation has a stark impact on his/her attainment. This is supported by the plot for London below that shows the relationship between a borough’s national deprivation rank (known as the index of multiple deprivation or IMD).

Another way to show represent this information is by mapping the 2010 GCSE scores for each of the London Boroughs and resizing the borough so that it represents the levels of child poverty (measured by number of under 16s receiving means-tested benefits).

Again, the map above is not perfect as it is still quite generalised and shows only one of the many measures of child poverty that are used. Both maps also show only one measure of attainment the “GCSE or Equivalent” score. The “or Equivalent” bit is important here as it covers a wide range of more vocational qualifications (called NVQs) that are often perceived as less academically challenging and can be a way for students to get the equivalent of 5 A* to C grades including English and maths (a key educational benchmark) without having to be proficient in these core subjects. This is important as schools in England are often ranked by the proportion of their students achieving this benchmark resulting in a possible bias towards the schools offering more vocational subjects and against those offering more challenging ones such as modern languages. It is interesting to consider whether the nature of equivalent qualifications makes them more likely to be used by certain types of school and to explore this further I have produced the plots below. The codes are as follows: AC= Academy, CTC= City Tech. College, CY= Community School, CYS= Community Special School, FD= Foundation School, FDS= Foundation Special School, IND= Registered Independent School, INDSS= Independent Special School, NMSS=Non-Maintained Special School, VA= Voluntary Aided School, VC= Voluntary Controlled School (if you are as baffled about these as I was see here or here).

The plot shows 9 regions of England. Each point represents a school in that region and is coloured by its type. On the x-axis is the inverse (higher= better) regional ranking of the school based on its GCSE scores only and on the y-axis is the regional ranking if “equivalents” are included. If the inclusion/ exclusion of  equivalents made no difference to the rankings then the points would follow the grey lines perfectly. In reality we get schools falling either side of this line with those under it benefitting if equivalents are counted and those above benefitting if they are excluded. For example, broadly speaking independent schools (light blue) look worse when GCSE equivalents are used in the ranking criteria and therefore would benefit if such qualifications were excluded. This also seems to be the case for the voluntary controlled schools in pink. Academy Schools (orange) however do much better with the inclusion of equivalent qualifications perhaps reflecting a more vocational emphasis to their curriculum. There are also some interesting regional distinctions with independent schools, for example, in the South West and South East appearing to do well whatever the ranking criteria whilst the East/ West Midlands and the North East present a more mixed picture. I think a lot more can be said about these plots so I would welcome comments!

Buried in the London Datastore are the population estimates for each of the London Boroughs between 2001 – 2030. They predict a declining population for most boroughs with the exception of a few to the east. I was surprised by this general decline and also the numbers involved- I expected larger changes from one year to the next. I think this is because my perception of migration is of the volume of people moving rather than the net effects on the baseline population of these movements. I don’t envy the GLA for making predictions so far into the future, but can understand why they have to do it (think how long it took initiate Crossrail!). Last year I produced a simple animation showing past changes in London’s population density (data) and it provides a nice comparison to the above. In total I have squeezed 40 maps on this page!

Technical Stuff

These maps were all produced to demonstrate the mapping capabilities of R. The first uses ggplot2 (plus classInt + RColorBrewer) and is based on some code (see below) written by Mark Bulling. If you follow the code below you will end up with this map, not the one I have produced above. I will stick my code in a formal tutorial soon. The animation uses the standard plot functions (plus spatial packages) in R as per this example.

The TFL data release contained the start point, end point, and duration for around 1.4 million bike journeys. An educated guess has been made about routes between stations using OpenStreetMap data and some routing software. The animation shows the scheme’s busiest day (thanks to a tube strike) and provides an amazing insight into the dynamics of Boris Bike users. You can find more info here.

I suspect this animation will be another big PR win for TFL, it is just a shame that it took a freedom of information request to get the underlying data.

Martin’s viz is one of my favourites but there have been a couple of others released that use similar technologies to show urban transport systems. Chris McDowall has produced an animation on a much larger scale by showing Auckland’s public transport system on a typical Monday.

Another great animation was produced by fellow CASA researcher Anil Bawa-Cavia. This shows London’s bus network and it makes for a great comparison to Auckland’s transport system above.

Inspired by the What’s in a Surname? map we helped make with the National Geographic, I have created 15 interactive typographic maps to show the most popular surnames across London. What they lack in cartographic brilliance, I hope they make up for in detail. There are 983 geographic units (Middle Super Output Areas) in each map and across all 15 there are 2379 individual surnames (15,000 surname labels in total). The font size for each surname label has been scaled to give an idea of the number of people who have that surname in each place. The surname frequencies come from the 2001 Electoral Roll and won’t contain everyone living in London but it is one of the best datasets available.

London is renowned for being a diverse city but this is barely reflected in the most prevalent surnames- only a few name origins can be discerned from the map. You have to look a little further down the surname rankings for this diversity to become apparent. The surnames shown on all 15 maps can be traced back to one of 38 origins; I have selected unique colours for 10 of the most popular. Surname origins were established using the Onomap classification tool. We are mapping the origins of the surnames, which are not necessarily the same as the origins of the people possessing them. Many people in London have adopted Anglicised surnames.

It is also clear from the maps that the same sorts of surnames tend to cluster together. This is because they often closely reflect the naming preferences of particular groups of people within an area. As you transition through to the less popular surnames things become a little more jumbled and the distinct patterns present in the first map become less distinct.

The final thing that stands out is how surname popularity decreases between the first and second most popular names and every subsequent change after that. You can see this by how quickly the text size reduces until almost all names are written in the smallest font sizes.

The more you study these maps the more interesting, and perhaps complex, they become.  My final thoughts therefore appear a little contradictory. The first is that a surprising number of Londoners share the same name (especially with their immediate neighbours). The second is that despite the dominance of relatively few surnames at the top of the rankings, the further down the rankings you get the more you see of London’s population diversity. We are of course only mapping the top 15 surnames in each area of London- there are many thousands more. If you can’t find your surname on these maps, you can see where it is around the world here.

The maps were created as part of my ongoing PhD research using the Worldnames Database compiled by University College London’s Department of Geography. Thanks to Oliver O’Brien from CASA for putting the maps online. A high resolution print version of the map (previewed below) is available on request.

The visualisation above shows the average relative duration of Boris Bikers’ weekday journeys over a 4 month period at hourly intervals. For each time step the average journey time (in seconds) from each docking station has been calculated.This information is interesting because it shows the preference for short journeys around the City of London, whilst people on the outskirts of the the scheme (especially to the west) take longer journeys. I also like the the fact that journey times around Soho and the West End are longest around 23:00- perhaps correlating with the number of after-work drinks consumed. In one visualisation you get to see the changes in the cyclists behaviour- from the early morning commuters through to the late night cruisers

The data come from Transport for London’s recent release of 1.4 million Barclays Cycle Hire journeys to their developers area (thanks to this FOI request). The data are said include all the journeys between 30 July 2010 and 3 November 2010, except those starting between midnight and 6am. In this analysis journeys taking more than one hour are not included (there are relatively few and many were actually the bikes being removed for maintenance) and docking stations with fewer than 10 journeys within each hour across the time period have also been ignored.

The maps can be improved in many ways- stay tuned for more developments and I will also post something a bit more technical about the methods I used etc to create the map (I used a strange cocktail of R and ArcGIS 10) .

I also recommend Ollie O’Brien’s (@oobr) brilliant interactive visualisations these data.