#tableau dataviz ironviz | Explore Tumblr posts and blogs

dustydata · 8 years ago

Text

Directing Darwin - 2017 Iron Viz Feeder 2

It’s been a while since I last posted, but i couldn’t say no to another shot at the Iron Viz competition! The second feeder of this year had a plant or animal theme. I really like having themes that are not centered around new functionality related to data sources. Specifically, the shapefile challenge last feeder was EXTREMELY difficult to find good data to utilize. Not to say that finding data this time was easy (data is always always the hardest thing in these competitions), but I was pleased to have such a broad topic to choose from.

1) Find the Data

I went through various ideas for this feeder including the below:

- Coral Reefs

- BP Oil Spill

- Mythical Creatures

- Deforestation

- Migration Patterns

- Darwin, Mendel, Pavlov

Eventually, I settled on a story that i had seen on an online video once before. One of the keys to winning an iron viz is the story. I know that the judging says that its only worth 25%, but in my opinion its closer to 70%. Reason being, when someone is interacting with your viz, they don’t want to have to figure out why they should pay attention. Just like a good book or movie, the story should grab you immediately so you CAN’T look away!

Taking the logic above into account, I chose the topic of the Russian Fox-Farm experiment. The 2 minute version is that during the late 50′s a Russian scientist started a study to determine if and how domestication could be influenced by selective breeding specifically tailored towards “tameness” traits. After 50 years the experiment had yielded almost completely domesticated animals which closely resembled dogs in personality and visual traits.

2) The Build

For the build I knew that I wanted to have a dashboard that was long and scrolling. Honestly, one of the main reasons for this was because I knew that I would have so much content that it would be incredibly hard to pack it into anything besides a 4000 pixel long dashboard! After some tinkering I figured that my dashboard would be split into two main pieces. The first section would be dedicated to explaining what domestication was and which animals were domesticated. The data around the fox experiment itself was fairly sparse and I felt that having more robust data to investigate would be great for background info and functionality

I wanted to keep the visuals very clean for this dashboard and kept the colors to 3 (yeah three colors) for the entirety of the dashboard. It’s the least amount of color variation i’ve used in a dashboard, but I tend to get lost in dashboards that have too much color and I really wanted to direct users’ attention to the most important pieces of the viz!

The middle of the dashboard is the most dynamic functionality-wise. It breaks down the taxonomy groups of animals who are either fully domesticated or semi-domesticated. The user has the ability to click into each portion of the treemap (sized by number of animals) and it will impact the full listing of animals (in timeline form) on the bottom left. This listing will change the picture displayed by hovering over the line. When you hover over the box to the right it will give you more information including why the animals were domesticated and where they come from.

The meat of the story comes from the Fox Timeline which has points running through the last 50 years of the experiment. The key point of design here was the double helix. This took me a little bit to determine how I was going to accomplish it, but in the end it wasn’t too bad! The basic steps were:

1- Find png of double helix

2- Stack them together in powerpoint, save as image

3- Import as background image, annotate points, save coordinates in excel

4- Map points background shape

The timeline continues and gives various details about the fox farm experiment. I tried to vary between charts and visuals to keep the reader entertained without “mathing” them to death.

The final visually interactive piece of the dashboard is the diagrams of the foxes. Obviously, everyone interested in the story to this point wanted to see what the foxes looked like after domestication. I found some really great drawings online of what the foxes looked like before and after the study and it was stunning. You can really see the transformation into a dog-like fox! Originally, I thought about having traditional bubbles to annotate the visuals, but in the theme of keeping it simple, I kept the icons as simple shapes with tool tips that give further detail.

The final touch was a conclusion paragraph to wrap everything up! All and all, I feel really good about what I created and I hope its an interesting story that everyone can gain some insight from!

Best of luck Iron Viz competitors! (especially my wife who is competing for the first time in this feeder!!!)

#ironviz tableau dataviz

5 notes · View notes

dustydata · 9 years ago

Text

Mr. Popularity

Tableau’s second feeder contest for the awesome IRON VIZ competition at the conference was announced on June 10th! I figured that since I missed the kickoff for the 1st feeder (food related) that I would give this one a shot! Now, I’m sure that I could have entered my “Presidential Prose” dashboard (pic below - direct link), but I thought that I needed to create something new just for this contest!

Step 1: Think of an Idea

My first task was to think of an idea....and really that is the one of the toughest things to do when creating one of these dashboards, besides actually finding the data. I went through a lot of different ideas - women in politics, highlighting the LOSERS of presidential elections, lobbying money, etc. What I finally landed on was Presidential Approval Ratings. I thought it would be a near universal subject that wasn’t tilted one way or the other and that many people could appreciate.

Step 2: Find a Data Set

Finding data, at least to me, is the hardest part of a contest like this. What data is right? Where do I get it? Has it been used already? Is it clean? Is it relevant? After hours of searching the google machine (as well as my soul) I found a decent dataset from The American Presidency Project. The data wasn’t super-pretty but it was manageable (see below).

After a few attempts, I was able to scrape the data from the site for all the presidents back through 1941! It was simple data, just a few fields wide (Dates, Approval, Disapproval, and President), but that’s what I wanted. Too often I find myself trying to find the most complicated data, when in reality it’s often the simplest data that is the most impactful!

After thinking about what I wanted to do, I decided that I wanted to have a timeline and some key images of the president. Almost immediately I also thought.....no that’s too simple, what are people going to want to know?? I can’t just give numbers. So back to the google machine I went and then I realized what I needed ------ events! If people saw that a president’s approval was high or low, they want to know WHY. Now, you can never truly know all the factors that make up a single point of data like approval, but national events shape approval more than anything else, so that’s the direction I went.

After a few more hours, I realized again that I was overthinking things. What do I use ALL THE TIME to answer my questions about...well...everything? That’s right! Wikipedia! It was the “you can’t use that as a source” site that I knew in school anymore. It actually was considered to be more accurate than most textbooks! And so I found a gem of a page called “Timeline of United States History”. It wasn’t perfectly formatted, but it had all US major events dating back to 1513!

After scraping the data, layering on some other presidential info (# of president, term, party, etc.) my excel file looked like the below.

Ok, so I’ll fast forward through the next details of working and re-working my dashboard until it met my expectations. Below is the initial view of the dashboard that I ended up with.

What I like about the view is that its split into a few neat pieces. The top shows all of the presidents in the data set and their images. Underneath that is a neat chart that I don’t even know what to call (lolipopish?). It shows the max, min, and average (the circle) of the president’s term for approval. Each president has their own corresponding color, which matches the bottom chart. If you click on a picture of the president, it will show a little summary of the president and their highest approval rating / event. I thought this would add some nice detail without being overwhelming.

The bottom chart shows each president’s approval rating by week from the dataset. President Obama actually had daily approval ratings in the dataset, but I used a weekly average so that the total average for all presidents wasn’t skewed. The initial view shows each week as a circle. If you hover over the circles it will show more detail and if a historical event (from the wikipedia dataset) happened that week, it will show up. One cool example is President Truman. I set up the chart to show which points were outliers (3 standard deviations) and one point in particular popped out. In the screenshot below you can see that Truman had an astounding 91% approval rating the week that the Atomic Bombs were dropped on Hiroshima & Nagasaki. This was just one neat insight among others that I found (Nixon’s resignation, 9/11, etc.).

The other piece of functionality that I added was to be able to view the dataset of approvals vs. disapproval in a scatter. I added a parameter to be able to flip back and forth from these views. Below is what the scatter looks like.

I really enjoyed making this viz. It’s neat when you look at your own dashboard and end up playing with it for a few hours finding tidbits of interesting history! Good luck to all the other viz feeder contestants!

Oh, almost forgot - here is a link to my entry!

http://tabsoft.co/1ZRxwpE

#tableau dataviz ironviz

0 notes