The Walking Viz? The Wiki Dead?

In preparation of the 2015 Iron Viz competition, Tableau Public is hosting their first qualifier. The first challenge is to create a Viz from Wikipedia data.

This is my first attempt to participate in this type of Tableau challenge, and I knew going into it that I have a lot to learn. It is pretty clear to me that this won’t be winning material, but I’m okay with it being good practice.

wikidead_dashboard


The Approach

I’ve been traveling on the train a lot lately, and was catching up on the Walking Dead. As I toggled over to Wikipedia after watching an episode, it was the first thing that came to mind. I started searching and I found a number of interesting links.

  • http://en.wikipedia.org/wiki/The_Walking_Dead_(TV_series)
  • http://en.wikipedia.org/wiki/List_of_The_Walking_Dead_(TV_series)_characters
  • http://en.wikipedia.org/wiki/List_of_The_Walking_Dead_episodes

    There were also other links of interest, including a page for each season and a list of all awards that the show was nominated for. With the show information, it looked like I needed to do text analytics to get any real meaning out of those pages. I spent some time trying to manually clean up the awards section, but ultimately it was taking too much time and I wasn’t sure how it was going to enhance my analysis. Perhaps comparing millions of viewers to awards… but I was running out of time.

    Ultimately I stuck to the main data presented in the Overall series, characters and episodes. It took me a while to convert all of the color notations to real data elements in my table. I suspect that there is an automated way to do this better, but given the limited amount of information I used the brute force approach. It did cost me a lot of data prep time.

    The Design

    I was having a lot of fun with some of the design elements, when taking breaks from my data manipulation. I found a comparable font to The Walking Dead title, and started in on making my own icons of zombie heads in PowerPoint (my corporate design tool of choice).

    zombie heads

    From there I sampled colors off of a promo image from the latest season. I ended up tweaking these a bit because I needed more than the 5 colors derived.

    download

    Once I had finally finished all of my data manipulation, I found my actual data to be very limited. I would have loved to have known how many zombies each character had killed by episode for example. I found it on another site, but not Wikipedia. While I’m not really a big fan of the packed bubbles, I did like the idea that it was like looking at a group of people. Similar to looking at each of the characters, you have a sense as to how well they will fair. Yet, someone like Beth Greene who lacked Daryl’s grit, was in the top 10 of episodes survived for the entire season.

    The Analysis

    Ultimately I was hoping to be able to predict when any character might die. One of the biggest issues that I had with that is that a character’s role (main vs. recurring) and the various levels of credit billings could change between seasons. From this data I also wasn’t sure if it changed between episodes. Until I spent the time to bridge that gap, I didn’t feel that I could make an accurate assessment.

    character_distribution

    This distribution didn’t make it into my viz, but was interesting. You can see that there are some clear outliers by role, but otherwise there are pretty tight bands around character survival rates. Even within main characters, you can see that if you aren’t one of the first survivors, your chances aren’t that great. Given more time, I would have looked at this data by “highest” level of credits achieved. Although this wouldn’t tell you if that new character is going to make it or not. The guest roles are clearly limited however, and I think that those are mostly cameos after a character has already been eliminated.

    The ratings alone by season and episode are certainly very interesting. Obviously the season premier has much higher viewership and that starts to drop off significantly. By having the current seasons broken into two parts, you can see the impact of bringing those viewer numbers back up. Beyond that, while we have seen significant growth in the show’s popularity, it looks like that interest is starting to level off. Maybe it’s time for Carl to head out on his own for a spin-off.

    BOTTOM LINE: Stay close to Rick or Daryl if you want to survive the zombie apocalypse!

    Challenges

    My biggest challenge was that I should have interrogated the data further before choosing the topic. Since I split my time between design elements and data from the beginning, I was too committed to turn back. I ultimately found a much more robust data set, but didn’t have enough time to work with it. It looks like I’ll hold onto that one for a future competition.

    I also wanted to make my own Terminus map in Tableau using the Drawing Tool from Interworks, but that was very time consuming and was never completed.

    The Results

    Here is my working Visualization. It was certainly good practice and I look forward to the next competition.