Presenting data in new ways: Horizon Charts

As it becomes easier to access an increasingly broad range of publicly-available data, one of the challenges for us within the Observatory is to try and help make sense of large datasets and present statistics in a way that communicates key messages effectively.  This blog article presents an alternative way of illustrating large volumes of trend data, using unemployment to demonstrate the approach.  We’d appreciate your views on whether this is a useful technique.

Unemployment is one of the key datasets we monitor in the Observatory and our analysis of the trends (often at the very local level) feeds into a wide range of needs assessment work.  The volume of data on this topic is huge; figures are released every month, broken down into age groups, sex, length of unemployment and all available as both volumes and rates.

On occasions there is a need to examine large quantities of this data simultaneously and it can be difficult to visualise the figures in a useful way.  For example, we have 30 Localities across the county.  If we were to try and compare trends across these areas, the default approach would be to produce a chart along the lines of the one below.  We have 30 Localities x 12 months x 7½ years, or 2,730 pieces of data to understand.

Hopefully you’d agree this is pretty unintelligible.  There is far too much data to make sense of it in this format; it is hard to tell which line is which, how individual lines fluctuate and how they compare to the overall county trend.

An alternative method is to use something called a Horizon Chart. Take a look at the example below and see if you can more easily pick out the key trends.  Although it would be simple to pick out some statements regarding the best or worst performing areas, there are often occasions when people have an interest in a specific geographical area.  This can be excluded when only the extreme values are looked at, so this approach helps summarise the complete dataset.

Rather than present the precise values for every single month for each Locality (this can be made available via a spreadsheet if that level of detail is really needed), this chart focuses instead on key patterns and underlying trends.  It presents the degree to which each Locality varies from the overall county unemployment rate.  Basically, the darker the shade of red, the further away the unemployment rate is from the county average on the high side.  The darker the shade of blue, the further away the rate is away from the county average on the low side.

Click on the image to see the full-sized version.

The chart identifies some important messages that are harder to pick out from the first version.

  • Inequalities have increased significantly during the economic downturn.  Look at how the darker shades appear more prominently in early 2009.  This means that the areas with the highest unemployment rates are moving further away from the county average.  Throughout 2009, 2010, 2011 and into 2012, the Abbey & Wem Brook Locality area has had an unemployment rate typically around 4 or 5 percentage points above the average.
  • Conversely, the emergence of the darker shades of blue at the same time demonstrates that the Localities with the lowest unemployment rates in the county have become even further away from the average.  The gap has widened.
  • There is also relatively little  movement above or below the county average; if a Locality has traditionally had an unemployment rate below the average it is almost certain to remain that way into the future, and vice versa.
  • We also see that four of the five highest unemployment rates are found in Nuneaton & Bedworth Borough.  The Localities have been sorted in terms of how far away they are from the county unemployment rate, on average over the period in question.
  • However, it is too crude to regard the issue as a straightforward north/south divide.  We see that other parts of Nuneaton & Bedworth Borough, namely Weddington & St. Nicolas and Whitestone & Bulkington, are consistently below the county average during the past eight years.

Although it may take some getting used to at first, this approach could work quite well.  Let us know what you think.

3 Responses

  1. Lovely job, think it’s really effective, any chance you could share the methodology?

  2. […] Presenting data in new ways: Horizon Charts […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: