It is time for the first IronViz feeder! The topic: Agriculture. The mandatory dataset: 2012 Census of Agriculture.
My entry is > FARMSTATES
THE INSPIRATION
FarmVille was a fever when it was first released and I believe the reason a lot of people migrated to Facebook from previous social media platforms. I know that in my country Brazil, a lot of folks migrated from Orkut to Facebook to build and keep their farms.
When I first saw the dataset announcement I thought with myself… can I do something that recall FarmVille? And that’s where I got the title of the viz: FarmStates. Now, I wanted to use a similar font for the title than the one used in FarmVille. By Googling FarmVille font I landed in a very nice page: https://fontmeme.com/farmville-font// The Font Meme website provides hundreds of free fonts and, better, it allows you to write your text, select colors and special formatting and… download a high resolution image of your custom font WITH a transparent background. I will be using Font Meme to build custom text for Tableau visualizations.
Here is the original FarmVille logo:
Here is FarmStates:
THE DESIGN (MAP)
As many of the entries to this IronViz feeder can confirm this dataset screams for a map. So I had to design a map that could be both informative and also beautiful. That’s when I decided to build a double-axis map, one axis using the recently released Density mark and another more traditional using the Circle mark. As I think this is kind of a new design, let me explain how it works.
1 | Drag State and County fields to Detail. This will generate the Circle map
2 | For the sake of this example, exclude the Hawaii locations
3 | Change the mark type to Density
4 | I’ve picked the Density Multi-color light color option and set it to 60% intensity and 70% opacity
5 | Formatted the map layers by removing all but the “Coastline” layer. This makes a very clean map that servers as background for the data I am plotting in
6 | Created a duplicated Latitude pill on the Rows shelf (by CTRL+Dragging the existent pill)
7 | Right clicked the second pill and selected Dual axis. That will plot the two maps on top of each other.
8 | On the first marks card, changed the type to Circle.
9 | My viz is all about counting farms by County so I’ve created a Calculated field to build a specific measure for it:
IF [KPI Description] = “Number of Farms: 2012” AND [State] <> “Total US”
THEN [Value Numeric]
END
10 | Dragged the new field (Number of Farms) to size card on the Circle map.
11 | Clicked on Color card and made the dots white with a dark gray border.
12 | Double clicked on the Size legend to adjust the size of the min and max bubbles.
13 | Edited tooltips and added color legend by using annotations. This was required once I had a sheet swap action to change the map from County to State level (hex map). The map ended up like this:
The benefits that I see on this design over the plain density mark is that I can see another metric on the map as the size of the bubble represents the quantity of farms in each County. Also it allows me to identify where these Counties are and make it easier for the tooltip interactivity to happen.
THE ANALYSIS
The visualization is all about the Farms themselves. From where they are, to the gender diversity of their main operator, I’ve detected the indicators in the dataset that could be more interesting for any reader to become more knowledgeable on the farms industry in the US. The second part of the visualization is exactly an analysis of the farms and their respective Counties population. By using the Clustering features from Tableau I was able to detect 3 main groups (I am not sure general public understand what a Cluster is). Then it is interesting to find that 87% of all farms in the US are located in Counties that average a population of less than 100K people.
On the Scatterplot I also used some techniques from Pooja Ghandi that generates a radar-like effect and highlight the values on both axis. This is built by adding two Reference Lines and set them to the MIN value of each axis. Then make sure to select “Show recalculated line for highlighted or selected data points” and, finally, add a Dashboard Highlight action on the Scatterplot. I’ve called it “Radar” action.
Finally, I inherited the colors from the density map into the groups to create a visual identity on the viz. Later on I got a very valuable feedback from Sarah Bartlett to adjust the colors of the title to also match the ones from the density map. Thanks Sarah!
THE STORYTELLING
I tried to my best to guide the user through the story of these Farms. From where they are, how to they group by population, what they produce (crops and animals) and… booo… which ones use Chemicals on their crops. And that’s when the visualization turns dark and focus on the issues caused by the use of Chemicals in farms, how they can be bad for the environment and for you. And to show that in practice to the reader, the climax of the visualization is the exact same map we have at the beginning, but now coloring the dots in dark gray and sizing them with the % of expenditure of the Counties farms on Chemicals. The original beautiful map with pastel colors is now marked dramatically with tons of black dots showing where farms are expending a lot on Chemicals… and where you, as the reader, should avoid buying food from!
To wind down the story I then offer some advise on what the reader can do in order to protect their families from the CHEMFARMS and also to change this scenario by demanding public policies that favor farms that are friendly with the environment. If I want the reader to take anything out of this vis, those last 5 bullets are them!
So a bucolic visualization named FarmStates then wraps up with a message and something the reader can reflect on. ChemFarms are dangerous and we can make an impact by changing our habits and demanding those public polices changes from our officials.
THE MOBILE VERSION
Some years ago the “mobile first” was the bible for people designing content to the internet. With the advances on response design implementations in CMS most of new web pages are being born, from scratch, as fully responsive to any device. It was my goal from the beginning that my IronViz entry would work nicely also on mobile devices, in special because the announcement of its release will be via Twitter, a platform that people consume mostly from their phones. So I didn’t want people to click on my viz and then not being able to see anything because it wasn’t optimized for their device. FarmStates IS optimized!
The Process: Tableau does offer us the possibility to build a design that will work fine for small devices. And this feature was enhanced recently as Tableau tries to its best to generate an automatic phone design to every visualization. Although it is not perfect, it is a way to start building the responsive viz. Due to the complexity of some charts I had to sometimes duplicate them and create a (Mobile) version of the same chart. In special for those that contain annotations. I also had to reduce the text size by half in all text boxes so that they could fit and be in a good size for reading in a mobile phone. I can say that it took less than one hour to have the viz 95% adjusted to the phone. And I believe it just look amazing there!
THE THANKS
Besides Sarah who reached out and provided feedback, some #datafam friends also helped a lot on building FarmStates. So the Thank You note goes to the amazing Ken Flerlage , Bridget Cogley . Lindsay Betzendahl . Lindsey Poulter and, finally, Alex Waleczek for the Tableau Prep work.
Tableauing 