It was a great class not only for those to whom the craft of Data Visualisation was new to them but also had some great nuggets for those wanting to optimize their design approach.
As part of the class there were practical activities and one of note was to explore a data set on obesity trends from theWorld Health Authority. Delegates were asked to look at the dataset in excel and identify data variables and ranges, ways they might transform the data and crucially identify some possible data questions and interrogations. Andy then shared his answers and his data visualisations with the class. This activity got me thinking ….
Could I use SAP Lumira to undertake these tasks and recreate the data visualisations ?
Andy kindly shared the dataset with all the delegates and on my train journey home I started to play. I acquired the data set into SAP Lumira which was only 11,580 rows and 14 columns. Not a large dataset by any means but packed with insights to be drawn out. I then set out to reproduce each of the visualisations discussed in the class using SAP Lumira.
Not too difficult but I had to reduce the number of data items shown to the top 50 to allow me so see and therefore select (highlight) the USA ans UK
Now this was the toughest one in the set as SAP Lumira doesn’t have this chart component type in it’s library (an opportunity for an enterprising developer using the SDK maybe). So I worked through a number of different ideas but sadly had to pivot the data outside of SAP Lumira in Excel and add multiple data sets to structure the data in a shape to drive the visualisations I had in mind. This method worked easily and I will repeat this approach in the future.
Not as easy to read as the original
You know I just love Pie charts !! This kind of works but the scale doesn’t expose the differences well enough.
My thinking turned a corner … In Andy’s visualisation you are comparing the gap between the BMI in each gender by region, you could call this the variancebetween the gender scores. So I set to work on showing the variance between the gender as an absolute number and then plotting how it changes by Region and between 1980 and 2009.
You can easily see the changes in the variance for example in Africa between 1980 and 2009 has widened by nearly 1 full point. This approach worked but it masked the gender split.
Then again my thinking moved on to using a Radar Chart which I think is the clearest and closest to the original representation by Andy Kirk.
Visualisations 4 & 5
These seams to be a “bug” in Asc and Desc sorts in SP13 but the visualisation are pretty much there.
Visualisations 6 & 7
Now this didn’t look tough until you think about the volume of data points plotted. 193 countries x 30 years = 5790 in a 6 zone trellis chart. Sadly SAP Lumira couldn’t render the chart.
BUT ….. If you can build one for one region then there is further possibility:
Use the new COMPOSE feature to build a storyboard with all 5 regions displayed.
It took me about 3 hours to prepare, explore the data and build all the visualisations and I’m really happy with the results. With more time refining the titles, colours etc. I think SAP lumira could really step up to the mark in delivering high quality Data Visualisations.
Content reproduced with the kind permission of Andy Kirk, visualisation blogger, designer, consultant, author, teacher, trainer and speaker