Welcome to #DuoDare. The Data Duo is back at it with a little friendly competition. Each month, one of us will pick a data source and dare the other to viz it better. We will pass our visualizations to each other for some feedback and present them to the community (you) to be the judge. Please fill out the Google Form to vote for your favorite duo viz. Our bragging rights are on the line so we need you more than ever!
Don’t forget, your monthly vote gets you an entry (up to 12) in our drawing at the end of the year for some free swag!
I got destroyed in month 1, but rebounded pretty hard in month 2. We should do more live voting in the future (did you miss our Tableau Conference talk, watch it here)!
I decided to look for some data to help keep the momentum going for Team Adam. I know geo data isn’t Pooja’s strong suit, so when I came across this UFO Sightings dataset from Kaggle, I knew I was done looking. I downloaded the scrubbed.csv, but I noticed there were still a lot of nulls everywhere. Also, while most records had a latitude/longitude, the location fields were pretty unreliable. They were generally a mix of abbreviations, spelling variations, and commas/parentheses galore. Welcome to the world of free-form data!
Anyway, I decided to clean the scrubbed version up a bit. I used QGIS to join the scrubbed.csv with a populated places shapefile from Natural Earth Data. I used the NNJOIN plugin to complete a nearest neighbor join. This essentially uses the lat/long fields from the ufo data and determines the nearest populated place from another source. It then joins a ton of location fields to the original dataset. I now had standardized location fields to use instead of the originals. I wrote a post on using QGIS to do the NNJOIN if you want to learn more. I exported the joined result as a shapefile and connected this to Tableau.
I had previously shared the csv file with Pooja, but since I am such a nice guy, I gave her the enhanced shapefile with the NNJOIN. Bonus points?
Pooja’s Reaction to the Data
The last month was way busy for me with all the traveling and the Tableau Conference. I also lost the #DuoDare 2 vote miserably! 🤦♀️ When Adam sent me the data, my first reaction was – well, damn! It’s mapping data. I am not too good with maps. If Kent Marten is reading this, I am sure he is laughing. Because he knows some inside stories 😄
But while I was in Asia, Andy Cotgreave introduced me in all the events there as a very ‘competitive’ person. So nonetheless, I had to take up the challenge.
The Tableau Public experience on these is not great. For the best experience, you might want to download these.
The Away Team: Pooja’s viz
Adam’s feedback: Pooja did a really great job with this considering she hates maps! She has a great layout, minimal use of color and #Poojatastic lines walking the reader from summary level data to select highlights and finally to lower-level details. She always designs great interactions too. Hovering over the area chart will highlight the sightings on the map, update the dynamic text to tell you how many sightings you are looking at, and my favorite is the ‘O’ in the ‘UFO’ title is a hidden donut chart that updates and the label appears on highlight to tell you what percent of total sightings are in the view.
Other highlights for me are the use of some png files (pretty sure she built these in power point with soft edges) behind several elements. This soft glow is very subtle, but helps a few elements really pop. Boy, I can’t wait to do some of this stuff directly in Tableau someday. I also think Poo did a great job formatting the maps. She removed a lot of the map layers and just went with country borders and coastlines. This really helps to just show the data and not the underlying map layers. Shading the map to match the background makes it seamless without needing something custom in mapbox.
One area for improvement would be not covering the area (see what I did there <—-) chart with the map. I know she was trying to make the map as large as possible to fit within the constraints of the blog size, but there is some hidden data underneath. I struggled with this exact same thing on mine so I know she probably tried a bunch of ways around it. Tableau give us transparent sheets please!
The Home Team: Adam’s viz
Pooja’s feedback: Adam was nice to offer the cleaned and enhanced dataset to me this month. Maybe he felt the need to help me with geo data? Just maybe! 😂 But thanks Adam, that was huge help this month considering I had no time this last month due to all the travel. You’re the best!
Adam always does a great job with geo data. I mean did you see his geo spatial IronViz? He knows what to do with maps! This month he did a great job showing the United States map as the center piece of the visual. Breakdown of UFO sightings by month, year, time and weekday is a great touch because it not lets see high level sightings by location but also by different intervals of time. The intensity of the sightings reported is very clear with this approach.
I especially like the ‘case file’ section because it lets the user know what the report said about that sighting. The big number of sightings reported uses a cool dual axis trick to have dual colors on the text. Jeff Shaeffer showed this trick with text at the Tableau Conference this year. But Adam uses that trick very smartly to dynamically change the text based on hover actions. Very cool indeed!
His design skills are well known so I need not say much about that. Great work, Adam!
PS: Adam does a wonderful (NOT) job setting up the plot like, ‘I had previously shared the csv file with Pooja, but since I am such a nice guy, I gave her the enhanced shapefile with the NNJOIN. Bonus points?’ He ALWAYS does this. Sympathy votes do not count! 😂 Be smart about voting, Tableau Community!
It’s up to you now, vote for your favorite viz below.
Voting for this month remains open until we get around to posting month 4.