<< Back

Behind the headlines – COVID testing blackspots and iGeolise TravelTime

Around a million people are more than one hour’s drive away from a COVID-19 regional testing centre

The Telegraph, (29/04/2020)

Today we’re digging into the analysis behind this headline. We’ll focus on how we generate travel time extents and calculate proximity to COVID-19 regional testing centres.

We’ll be using the TravelTime (iGeolise) isochrone macro, built by The Information Lab consultant Ben Moss.

To follow along at home, download the workflows here (you’ll need to visit TravelTime to get a trial API key).

What is travel time?

Travel time is a far more effective way of assessing proximity (“closeness”) in our environment. It accounts for the actual route travelled, the mode of transport and the time of day. Rather than grouping people into an “as the crow flies” radius of distance, we can create personalised results that better reflect the real world.

Radius around Florence
People are not crows
Travel time by bike
Travel time by bike

Telegraph journalist Ashley Kirk used the newly released iGeolise Isochrone macro in Alteryx to investigate the UK’s COVID-19 testing centres and how close they are to how much of the UK’s population. . His analysis suggested that over a million people live more than an hour’s drive from a testing centre (at off-peak time). Let’s take a look at how he did it.

Building the analytics pipeline

We’re going to be creating isochrones in Alteryx. Isochrones are lines drawn on a map connecting points that share the same value – the travel time from a “starting” location. The connected line forms a perimeter around the selected “start” location. The area within the perimeter reflects what is reachable within a particular time – 30 minutes’ drive, for example.

We’ll be requesting the information from the TravelTime API using their Alteryx macros. The macros package up all the processing and reshaping of the data coming directly from the API leaving us with a clean, ready to use dataset. This makes things easy – no configuring of download tools or parsing required.

Alteryx – create the isochrones

  • With a list of UK testing centre addresses (correct at April 28th) use the TravelTime geocoder to find their latitude and longitude
Configuring the TravelTime geocoder
Accessible interface and configuration guidance in the geocoder
  • Use the test centre latitude and longitude values as the starting points in the isochrone tool
  • Specify one (or many) travel times for the isochrone (in this case 15, 30, 45, and 60 minutes)
  • Select a mode of transport (driving, departing Sunday morning, off-peak)
Isochrone configuration
Drop-down and check-box configuration for the isochrone tool
  • Hit run
  • The flow takes about 30 seconds
15 minute drive time
15 minute drive time coverage
Complete coverage – 60 minute drive

Overlap with UK population

In order to assess how many people are outside a 60 minute drive (not just draw the isochrones), we need to know the UK population coverage. Ashley used population at 1km grid level. I’ve used the same data made available by the European Information Data Centre.

There are two ways we could calculate how many people live outside in the one-hour drive time areas:

  1. Find the remaining land area of the UK outside of 60 minutes’ drive and find its population
  2. Use the total isochrone area, find its population, subtract it from the total UK population

I’ve gone for option 1 from an efficiency standpoint – the remaining land area outside 60 minutes’ drive is much smaller so requires less processing.

Population intersect in alteryx
Finding the population that lies outside the 60 minute drive

This is pretty heavy lifting in processing terms so you’ll have time to brew a coffee while you wait. At the end of the flow we are left with a number of people who are outside of a 60 minute travel time isochrone.

That number? 1.8 million.

Ashley’s article doesn’t specify more granularity than “over one million”. It’s highly likely that our numbers won’t match exactly due to:

  • Precise configuration of the macro (choice of day, choice of time)
  • Exact method used in disaggregation of 1km population values to smaller grid sliver areas

However, the methodology stands up as a very strong broad identifier of the scale of the COVID-19 testing accessibility issues in the UK.

Visualising the output

By feeding the isochrones through to Tableau I have replicated the visual that featured in The Telegraph article.

Adi recreates The Telegraph visualisation using Tableauview here

Final thoughts

Designing a flow that makes these API calls, delivers, and reshapes the results could be prohibitive for many Alteryx users, but accessing and configuring the TravelTime macros was incredibly easy. I found that I was (largely) able to complete my analysis under the limit enforced by the trial key. This has a 2 week time limit, and is capped to 10 requests per day (plus 100 Geocoder requests per day)

Why not give this a shot yourself for another region?

Next steps?

To download the TravelTime macros, visit the Alteryx gallery.

To get your free key to get the macros running yourself, register on the TravelTime website.

To download our sample workflows, go to this location.

Feel free to engage with me (@AdiBop_) on Twitter to talk more about the TravelTime tools or all things spatial.

Adi McCrea

London, UK

Leave a Reply

Your email address will not be published. Required fields are marked *