Makeover Monday, 2017 #16

Week 16 and it’s back to some big data on EXASOL thanks to Eva and Johannes. This time over 750 million rows of GP practice prescribing data for the UK from 2010 – 2017. The nominal challenge to makeover some of the charts in a House of Commons research briefing based on the data.



Like most of the Makeover Monday community I didn’t find the graphs particularly exciting or engaging. but I did find the research briefing to be interesting and it covered some of the key questions I had in my mind about prescriptions as a topic, such as what was happening with antibiotic use.

Also like others I didn’t simply try to recreate the graphs in the briefing. Actually that’s not entirely true, I do tend to start these makeovers by recreating at least one of the graphs or data tables. Recreating something is a good way to check that I’m getting the right numbers and helps me to orientate myself to the data. For my actual makeover though I wanted to dive into some of the detail like many others. One of the early submissions by Adam Crahen identified the growth in prescriptions of Apixaban. I knew a bit about anticoagulants, so this felt like a good category to drill into:



The first step was to identify the main anticoagulants being prescribed. I picked these four as an obvious story seemed to emerge; the move away from the traditionally prescribed Warfarin to three new novel oral anticoagulants. A little insider knowledge made me include Dabigatran as it doesn’t really stand out from the crowd (the remaining grey lines) unless you look closely.

Next I sourced a reusable image relating to the vascular system. Now at this point I got pretty excited. I thought that I could blend the arteries, etc from the image into the lines in the graphs. So I had the timeline running down the viz and was trying to resize things to line up nicely. End result? It looked like terrible! Back to the drawing board and I’m glad that I did.

Instead I decided to use the image as a backdrop to some context (using the title and the number of anticoagulant items prescribed compared to the overall number of items prescribed) and for aspects of the colour pallet. A much cleaner result I think whilst still retaining some visual impact.

The final step was to source detail for the labels on the left (the story) and edit this down to fit. Reducing the text down to what I ended up with was probably the second hardest part of the process! If I were revisiting this visualisation I’d tie the story into the headline a little more by specifically mentioning heart attacks in the section on conditions. I would also try to pull in some antiplatelet prescription data as aspirin has a part to play in the story.

In terms of new things tried in Tableau, this week I went with an entirely floating layout for the dashboard to see if that helped with alignment variations between devices and Tableau Public. The approach of floating everything seems to have helped a bit, although I maybe should have split the story points into individual text objects so that they would better align to the aspects of the graphs that they referred to?

On a final note it was great to play with EXASOL again – a pleasure exploring and working with this large data set remotely.

My makeover is also available on Tableau Public.

Makeover Monday, 2017 #6

Great fun exploring 105 million rows of Chicago taxi data for #MakeoverMonday this week using the data underpinning this article. The full data set was provided on a hosted Exasol database, purported to be the fastest in-memory analytic database in the world (and it was pretty fast considering the amount of data I was querying from the opposite side of the world).

For my makeover I’ve tried to relate the data to the topic by placing key summary information into a taxi fare style sign - like you would see on the door of a Chicago taxi. That colour scheme is carried down to graphs showing the distribution of trip distances, durations and costs by year along with an approximated heat map of drop off points (without trying to plot millions of points!). The interactive version allows you to filter to specific pick up areas, allowing you to get a feel for the cost and drop offs for those traveling from O’Hare for example.

Chicago Taxi DashboardStatic image only for now until I figure out how to make a cut down extract for Tableau public!