2 Replies Latest reply on Jul 19, 2018 11:17 AM by Annabelle Rincon

    Iron viz 2018 - Feeder 2 - Health and Well being - Is sport is really good for me or should I eat more chocolate?

    Annabelle Rincon

      About the Ironviz :

       

      I like the Ironviz season, I feel so emulated and stimulated, that is probably my second favorite period of the year ( just after Christmas). If you are not familiar with the Ironviz concept, just a few words. Ironviz is a Tableau competition, where people around the World compete against each other, in a friendly way, by creating a fantastic viz on a pre-determined subject. There is 3 feeders during the year plus one exclusive for Europe.

       

      Check the detailled of the competition here:

      Iron Viz 2018 - The Schedule | Tableau Public

       

      and the great viz of every participants, and winners :

      Iron Viz "Books and Literature" Feeder - The Winners | Tableau Public

       

      I suggest to everybody to participate in the ironviz or makeovermonday because there are  good exercices to go into the world and dare to be yourself and show what you are capable of.

       

      Having said that, the subject of this feeder was Health and well being, the contest details reveals that we can also used the quantified self. Iron Viz Global 2018 Feeder 2 - Health and Well-being | Tableau Public

       

       

      The Story:

       

      After some reflexion, and knowing that I will never gather enough data on my personnal fit training (no connected watch and so little time), I decided to do a funny visualisation about if I should really exercice or if it will be better to stay on the couch. If I choose to use  an humorous tone, it is because I sincerely think that using humour is the only way to impact people. In the past, to open self awareness, fear and saddeness were too commonly used.

       

      I had my idea, so the following point was to develop my story a little further and then gather enough real data to build the viz. The arguments against Sport will be given by a nice Demon Rat and the arguments in favour will be given by an Angel Rat version of myself. Respectively I will use black when the Demon is speaking and white when the angel will convey healthy arguments. To be sure that point will be understood, I specified into the viz.

       

       

      Capture d’écran 2018-07-15 à 18.59.05.png

       

      After collecting the data, I used Tableau prep, to join several tables or transform them, which was very convenient.

       

      One other particularity, that I wanted to add in my viz, was a third dimension. Not only, I wanted to create a story, but I wanted you as part of it. That is why I created a lot of interactions, that you can use to answer and discover the machiavelic  Demon's plans not to do Sport.

       

      Capture d’écran 2018-07-17 à 20.30.52.png

       

       

       

       

       

      Charts :

       

      In term of displays, I used essentially bars charts, but also a revised pie chart, a chloropleth map and a heat map scatter plot.

       

      1- Why did I use a pie chart?

       

      First that is not a pie chart, but a bike wheel chart. just kidding... I used this representation, because i wanted to display a statistic that represent 29% of the total, which is a little more than the quarter of the circle, and easy to spot. And because I was talking about cycle accidents, I wanted to give to this pie chart a riding twist, so I superposed two round, one as a pie chart and the other one darker to give this bicycle wheel look.

       

       

      2- Analyse the correlation between two variables :

       

      To analyse successfully the correlations between Obesity and physical inactivity or with Food Insecurity, I draw a bivariate chropleth Map and a Scatter Plot heat map.

       

      A Bivariate Choropleth Maps is very useful when you want to compare geographically accross two differente attributes, for instance obesity and physical inactivity.

      For each attribute, I determined a High - Medium - Low Rate, which gives me 9 couples, across 2 attributes.

       

      Formulas :

      I created respectively the percentile given me all the observations divided in 3 equal parts, so the Adult obesity rate is determined like this :

       

      Obesity rate :

      if [Adult Obesity (% of pop)]<=

      { FIXED :PERCENTILE([Adult Obesity (% of pop)],0,33%)} then "Low"

      ELSEIF [Adult Obesity (% of pop)]<=

      { FIXED :PERCENTILE([Adult Obesity (% of pop)],0,66)} then "Medium"

      else "High"

      END

       

      You do the same for the inactivity rate, and to determine the color coding, you simply create a concatenation

       

      color :

      [inactivity rate] +"-"+ [Obesity rate]

       

       

      For the interactions, I added a second sheets with the legend, so instead of the traditional color legend, which is static, the user can now click on the colored squares to highlight the corresponding counties.

       

      traditional color legend

       

      Final version :

      Capture d’écran 2018-07-17 à 22.48.57.png

       

      For all the steps you can use theses two excellents blogs

      Creating a Multi-Color Choropleth Map | Tableau Software

      How to make effective bivariate choropleth maps with Tableau | Tableau Software

       

      and for the choice of colors, I recommend you to use the colors suggested by Dr Cythia Brewer : Generalized Set of Color Schemes, or to read this very good research paper, for color in general https://research.tableau.com/sites/default/files/Affective%20Color%20CHI%202017.pdf

       

       

      The "Heat Map - Scatter Plot" allows us to see the correlation between these two indicators, but also to determine where are concentrated the counties. I learnt this technique during tceurope2018 in a wonderful seminar lead by  Jade Le Van.

      If you cannot wait to the next Tableau release and the heatmap viz type, which will allow you to use color to understand density ( Dev squad wows Tableau crowd with heatmaps, vizzes in tooltips, and cows! | Tableau Software ), the technique below) is for you.

       

      It consists on attributing bins of 1% to the two variables, so you obtain for each county a couple (x,y) that you can plot. For each couple you count the number of counties and drag and drop in color, et voila. So simple and I will never thought about it.

       

      Nb counties per bin couple

      { FIXED [Food Insecure bin],[Adult Obesity bin]:COUNTD([County])}

       

       

       

       

      I tried to create as many surprises for the reader as possible, so he/she never lose attention, have fun and learn something on the process. Without spoiling you, you will learn which cities are the most attracting for me, and the reasons why, you will see a beautiful picture of my dog, and the best natural parks where to bike.

       

      But before leaving this blog, I would like to show you, how I integrated a URL action, that allows me to create an automatic tweet with a pre fixed message, for the moment no one discovered it or at least no one send it. But it is quite obvious

      Capture d’écran 2018-07-18 à 22.59.58.png

       

      A click on the blue bird, create the following tweet "@rinconannabelle I like your @tableaupublic submission for #ironviz, and I would like to know the best place to buy chocolate!"

       

      To do so, you just need to create an URL action in the dashboard and add the following url

       

      http://twitter.com/home?status=%40rinconannabelle%20I%20like%20your%20%40tableaupublic%20submission%20for%20%23ironviz%2C%20and%20I%20would%20like%20to%20know%20the%20best%20place%20to%20buy%20chocolate!

       

      which create the following tweet :

       

      Capture d’écran 2018-07-26 à 23.33.58.png

       

       

       

      One last thought, always check your viz on Tableau Public before a final publication, otherwise you can be disappointed. Some fonts are not installed on the browser creating some differences when published. For instance at home, I see it correctly both on Desktop than on Tableau Public. But at work the font is transformed like this :

       

      Tableau Public version :

      Capture d’écran 2018-07-26 à 23.31.38.png

       

      Desktop version :

       

      Capture d’écran 2018-07-26 à 23.26.48.png

       

       

      If I was expected the font issue, I wasn't expected that the smileys will not appear... it is a shame because I was so proud of finding this idea to include smileys on reference lines....

       

                                       Tableau Public sans smiley                                                                             Desktop avec smiley  

      Capture d’écran 2018-07-26 à 23.27.12.pngCapture d’écran 2018-07-26 à 23.26.31.png

      I hope you enjoy reading, as I enjoyed doing it...

       

      Link to my viz : Tableau Public