What is this post about?
LEO is back again as the topic of this #VisualisingHE blog post. Why? Tableau Zen Master Chris Love nicely pricked my attention with his tweet:
Charlie Hutcheson among many others have engaged with this great conversational starter and call out for some ‘real life’ examples of Tableau in the wild. For Charlie this ended his blog posting hiatus and encouraged him to write a great post Hiatus over thanks to #RealTableau along with sharing a flurry of superb examples of #RealTableau dashboards on Twitter.
For me, this prompted me to share one of my viz by day V’s viz by night Tableau endeavours. Often my #VisualisingHE vizzes are spawned from #Dataviz by day dashboards. #Dataviz by night explorations both deepen my understanding of a dataset and further my own data visualisation skills (often allowing my more creative streak to get an airing). The LEO data set is becoming my favourite of late, it is interesting and very topical in the current HE environment due to the introduction of TEF. The LEO dataset is public and therefore lends itself perfectly for this blog post about my flow from #RealTableau to a Funviz, showcasing vastly different approaches to data visualisation.
This blog post focuses on telling the story of the evolution of an idea, highlighting how much I value the #VisualisingHE project which encourages me to practice my Tableau skills and benefit from constructive critique.
A little intro to the dataset
In case you haven’t read my previous blog on Introducing LEO’s Graduate Earnings, the Longitudinal Education Outcomes (LEO) data enables us to know how much UK graduates of different subjects at different universities are earning now, either one, three or five years since graduating. It does this by linking up tax, benefits, and student loans data”. (WONKHE beginners guide to LEO).
#RealTableau to Funviz
#Dataviz by day
1. #RealTableau viz. This standard functional data visualisation is designed to give a user an overview of an single entity (HE provider) performances against a sector benchmark (median annual earnings post graduation), display the difference against the benchmark (£) and clearly highlight the subjects performing above and below the benchmark. (blue v’s grey). #barchart
Interact with the viz on Tableau Public:
LEO Provider vs GB Sector earnings_#RealTableau
2. #RealTableau viz. This visualisation provides insight into the distribution of earning differences in relation to benchmark earnings for a particular subject. The business wished to understand a providers positioning within the range of differences in earnings across the sector, the viz allows for the selection of a specific provider (displayed as a larger dot in the plot). #boxplot
Punch the time card and head for home.
Que #dataviz by night
3. Searching for a better vantage point to see the data. Whilst the established #RealTableau vizzes provide insight into the relative performances of a single HE provider against a benchmark (per subject area), I wanted to get a better sense of the whole sector in one visual, so I started to try a few things out and explore the data.
This working viz provides you with actual earnings, difference AND the provider names, yet still this only looked at a single subject area of study at a time. I really want to get an understanding of the differences in salaries by provider and region in one visual.
4. Looking for inspiration I posted a few visuals to my friends and fellow #VisualisingHE comrades to see if they had any bright ideas.
Dave posted back with a short and sweet…. “I’ve got an idea but it will mess with your head”… #radialbarchart
He was absolutely right, a lovely challenge and a chart type I haven’t yet tried in Tableau, this should be perfect for presenting a ‘difference’ in earnings. Slightly less functional than the bar (4) but much more fun.
5. OK I know why I haven’t done one before… Math! Luckily Rajeev Pandy has written a fantastic blog about it and Charlie Hucheson had done a great take-apart Tuesday on a Radial Bar chart which should, in theory, allow for some pretty crude blind method copying.
Evening 1: didn’t go so well – more reading required! Confirmation that I need to understand these calculations rather than simply employing the blind copy and paste method! #notabigloverofunderstandingmathbecauseimreallyjustamusicanandloverofstorytelling
Frustrations messaged to team #VisualisingHE, met with encouraging words and support! Cheers peeps!
Que a bit more googling and a little read of Charlie’s blog to see if I pick up any pointers as to what I really need to understand and what I can just gloss over!
6. Evening 2: and the light bulb moment…………The issue was not the calcs, but simply that when reshaping the data – duplicating it up adding a reference id of 1 and 0 (necessary for this chart type), tableau brought the 0 and 1 identifier in as a number, you need to make this a dimension not a measure. Once this was on the worksheet, I got magical fireworks going off. #Wine
Happy faces and gifs galore from #VisualisingHE
7. Understanding the calcs and working out what they are doing in the chart (inner and outer radius). #justplayingandmakingprettypictures
8. Utilising the ‘combine fields’ functionality in Tableau to get control of the lines and groupings of the data.
9. Had a bit of a play with sorting. Firstly I wished to see if I could bring to life Dave’s drawing (4) but also wondered if it would read better if the differences were sorted by +/- sector median. #seashellaccidentalviz
10. One big picture… Explorations in trying to present all the subject level data and the regional grouping splits together in one viz? #onebigpicture #fireworks
11. I love accidents! Whilst playing with the viz messing about changing the chart type I stumbled on changing the line chart to be a square, oh how pretty! #accidentalart #uselessbutpretty
12. Back on focus. Having settled on presenting the regional groups by use of colour, I set out on multiplying the sheets, to create a grid of radial bar charts in small multiples. I did need to unpick the ‘normalising calc’ in the viz to set it to a max difference for all the charts, so that they could be compared together in one viz.
Next steps: A message out to #VisualisingHE seeking thoughts and comments on what’s missing in the viz. And help sizing the viz.
13. The value of friends to critique your viz is something I never underestimate. Dave sketched out his thoughts on the need for a top section of the viz to allow me to annotate the viz, describe how to read the chart and allow for space for the interactive highlights and filters to have their space in the viz. Elena on hand for #Vizoftheday advice on longer form vizzes and feedback. #whatthisprojectisfor
14. Ideas and comments. (13) Brought to life in Tableau.
15. Funviz. The finished viz | LEO Provider vs GB Sector earnings_Infographic (click to interact with viz in Tableau Public)
Hope this post has been interesting, and you have enjoyed following my flow of ideas and iteration with this particular dataset #RealTableau to Funviz