“We might not be able to fix the economy. But we can design the city to give people dignity, to make them feel rich.”
Enrique Peñalosa

City Bike

Transportation surrounds us in every facet of life (Delbosc, 2012). Pavements lead us to our supermarkets, bikes and buses connect us to our family and friends, and road and train networks take us across vast stretches of land to see every corner of our nation. We have all become familiar with, and used to, the luxuries of modern transportation. As transportation evolves to the challenges of the 21st century, such as congestion and climate change, it is important to consider our existing usage of transport modes and how we may shift towards more sustainable ones in the future (Buehler, 2012 and Laverty et al., 2013).

Through analysing open data 123456, we identified areas of England and Wales that share similar characteristics. Such identification is an essential step towards successful policy interventions and financial investments aimed at increasing transport accessibility and sustainability (Collins and Chambers, 2005). Therefore, as a group of multidisciplinary urban analysts from various corners of the world, we wish to share with you the story of our analysis.

MSOA map of the UK

Spatial Scale

For this analysis, we used census data per middle layer super output area (MSOA). An MSOA is a geospatial statistic used in England and Wales that was created by the Office for National Statistics. It is used to understand small areas of the country. The 2011 census reports 7,201 MSOAs with a mean population of 7,200 people and a minimum population of 5,000. While in some cases it may be helpful to aggregate MSOA data into larger geographical areas, the below sections will demonstrate the issues of such aggregation for our analysis of transport characteristics. We start with aggregated, national level statistics and work our way down to the MSOA level to show that variations exist at all spatial scales.

WFH = Working from home
Click on a legend label to see how total share would be redistributed without that mode.

Total Travel Share for England And Wales

Considering England and Wales as a whole, travel to work is primarily driven by car usage, followed by walking and taking the bus. Sustainable transport, which includes bus, train, metro, walking, and cycling, only accounts for roughly 28% of total travel share. This suggests transportation in its current form is not prepared to help combat the climate crisis. However, given the disparities in transport access and socio-economic conditions across England and Wales, total figures don’t show the full story. The most obvious discrepancy is between rural and urban areas.


Urban
(5875 MSOAs)

Rural
(1326 MSOAs)

Comparing Urban and Rural areas

As seen in the pie charts above, both urban and rural populations are highly reliant on cars, although this is much more pronounced in rural areas. Taking the bus is more common in urban places, potentially due to the larger number of bus stops per MSOA. The share of people working from home is much larger in rural communities, but in light of the COVID-19 pandemic we expect this will increase in urban constituencies. Walking continues to be an important mode of transport for both geographies, perhaps demonstrating that people will choose to walk when they can.

But such a crude categorization fails to show the variation amongst urban areas. For example, the next section breaks down the transportation differences between London and Birmingham.


London
(978 MSOAs)

Birmingham
(135 MSOAs)

Comparing Urban Areas

Birmingham’s average car share is almost double London’s, but London has far more commuting by the Underground and train due to mass investment in infrastructure. This is highlighted by the fact that London has its vast Underground network, whereas Birmingham has only a single metro line. Thus, there are differences in accessibility to various transport modes between these two cities. This is indicative of variation that may exist amongst all cities in England and Wales.

However, even this level of disaggregation is insufficient as it ignores the variation that exists within cities. In other words, an MSOA in Birmingham may be more similar to an MSOA elsewhere than to another MSOA in Birmingham.


Gif of transport variation in London

Analysing Variation Within Cities

As an example, London may be better served in terms of public transport than any other area in England and Wales, but services in London are by no means uniformly distributed. The animation to the left shows notable nonconformity across London for several variables, such as bus stop density and car ownership, which are used to understand transport profiles. Is it possible that some areas of London have worse service than other urban areas in England and Wales? The answer to that question, and many others, lies in looking at the data at the MSOA level.



You can view these maps as photos, by clicking on the links → Bus mode share , Car Ownership , Bus stop density , Metro mode share & Tube density

MSOA Level Statistics

Having established the need to conduct our analysis at the MSOA level, the following sections of our story examine public transport characteristics at this scale.



Public Transport Access

Looking at public transport accessibility, bus stops are by far the most prevalent type of transport facility. There are over 270,000 bus stops across England and Wales, with only two MSOAs having no bus stops at all.

That being said, other public transportation modes are less commonplace. Only a fourth of MSOAs have train stations at all, showcasing an uneven distribution in public transportation infrastructure. This is even more apparent when looking at metro, tram and underground stations as there are only 899 these. Only eight percent of MSOAs have one.

Variations in Travel by Car

Different MSOAs rely on cars to varying degrees. Unsurprisingly, a decreased dependency on cars is commonly mirrored by a highly accessible public transport network. On the other hand, MSOAs lacking a public transport network tend to be very reliant on private vehicle travel.

Sustainable Travel

Some of the transport methods can be combined into the following categories: active transport (walking and cycling) and public transport (bus, train, and metro). Together, these constitute sustainable transport. We can see that while some MSOAs have low shares of sustainable transport, others have extremely high levels of public transport usage or active travel. It is useful to compare the high usage cases to the average use of each of these modes across England and Wales and to highlight that higher levels of sustainable transport usage are possible in some places.

Average Travel Time Per Transport Mode

Train

67 Minutes

Bus

28 Minutes

Car

12 Minutes

Given the wide range of transport accessibility and mode choice, it is worth acknowledging that those who choose different commute modes spend different amounts of time on their commute.

Identifying Transport Profiles

It is evident that there is considerable variation and heterogeneity in transport characteristics of MSOAs across England and Wales. The next step was to determine whether different groups of MSOAs with a similar transport profile could be identified. To do so, we used the transport characteristics discussed above, such as commuting mode shares, car ownership and travel time accessibility by different modes to represent the transport profile of an MSOA.

We used a clustering algorithm to group the MSOAs into distinct transport profiles (Shelton et al., 2006). Clustering algorithms group data points to maximize similarity within groups and minimize it between them. Our clustering revealed five distinct groups in our data:

Good train accessibility but car dependant

This cluster is composed of rural areas that surround land-locked urban areas. The MSOAs in this cluster are mainly in the center of England and Wales, compared to the rural areas in profile 2, which are on the outskirts. This cluster has the second best accessibility scores for all measured transport modes due to the central locations. The cluster benefits from being on train routes and has the second highest train usage, but that is the only mode of public transport that the MSOAs in this cluster are serviced by. As a result, the cluster is associated with high car ownership and usage, followed by train and walking.

Solely car dependant

This cluster is made up of rural areas far from the cities. The MSOAs have few public transport options and people depend on cars to move around. They have poor accessibility even by car, and this could be due to a lack of direct road and other connections between them and other parts of the country. The cluster is found on the periphery of profile 3, which is itself made up of coastal urban areas with poor accessibility.

Lack of accessibility across all transport modes

This cluster shows the third highest usage of bus, bicycle and walking to work, but has the lowest train usage, working from home (WFH) and all around accessibility. The most popular modes to travel to work are by car, by walking and bus, but the lack of accessibility across all modes and little train usage is the defining feature. This can be found in coastal towns and cities such as Newcastle, Cardiff and Blackpool, which might suggest the MSOAs are at the end of train lines and other transport networks and therefore lack external connectivity.

High public transport and good accessibility

The cluster is associated with high usage of public transport including the underground/metro/tram, train and bus. It is noted to have very good accessibility to all MSOAs through all transport modes. This cluster dominates London, but can also be found in the centre of some MSOAs in big cities like Manchester and Birmingham. The cluster suggests that the transport profile of London is different to the rest of the UK and can only otherwise be found in high accessibility centres of large cities.

Car reliant but high public transport

This cluster has high car usage but is notable for the comparitvely large number of people who use the bus and walk to work. These MSOAs also have a high degree of accessibility but the overall transport profile is more shifted towards cars than the previous cluster. This is found in large Urban areas across the UK such as Manchester and Birmingham, suggesting that the main difference between these and London is the degree of usage of public transport with the main difference occurring due to the lack of usage of an underground/metro/tram.

Accessibility by Transport Mode in the Profiles

As explained above, the clusters differ in their accessibility. To get a feeling of this variation, the bar charts above compare the average accessibility by transport mode across the different clusters.

Relationship between Transport Profiles and Demographics

As an additional step we began to explore how, if at all, transport profiles relate to demographic characteristics (Harris et al., 2007). We attempt to answer this through a classification analysis, which is explained in detail in our methodology (Titheridge et al., 2008). The classification analysis was performed using a Random Forest Classification algorithm, including factors which have been found to be related to transport usage like population density, net annual income, and mean age. 65% of MSOAs were able to be correctly classified by the demographic factors, with population density and net annual income being the most important factors relating to the identification of the transport profile.


classification tree

An example of one of the decision trees from the algorithm can be seen above. It shows us how the algorithm worked to predict which transport profile an MSOA belonged to. For example, transport profile 4 (high public transport and good accessibility) is associated with an income above £33,850 with the percentage of households owned being greater than 59.25%. In contrast, transport profile 1 (good train accessibility but car dependent) is associated with an income below £33,850 and a mean age of less than 38.25 years. The issue, however, is that this is one of many decision trees created by the algorithm. To get a better understanding of the result of the classification, it would be necessary to examine more trees. This would help inform us as to which demographic factors are related to the cluster outputs.

View our interactive map →

Further Reading

Back to top ↑