A Tableau-centric weekly blog about the viz making process, #datafam member interviews, Viz of the Week & entertainment for introverts (consisting of a music morsel & a binge bite).
Viz of the Week
Disclaimer: This is the 1st Viz of the Week I was lukewarm about at 1st glance. The color killed me a little and thought it wasn’t oomphy enough. The community’s reaction to it and my subsequent obsession over it won out vs. my initial reservations & I even made my own version.
Adam Mico (AM): What gave you the idea to create a (Marty) McFly viz?
Zach Bowders (ZB): I’ve been thinking about the idea of novel timelines for a while, to the point where I decided to do a series of them as new ideas came to me. “Back to the Future” is one of my favorite film series and its chronology is fascinating. Much like with the movie Inception, where I really had to wrap my head around the mechanics of the dreamscape and storytelling, “Back to the Future” was a unique challenge. How do I represent the same time periods multiple times? Do I represent them as a single point or multiple, different versions of the same point since none of them are actually the exact same.
I’m definitely not the first or last to tackle this topic. I’ve seen two primary methods of representing the Marty McFly timeline.
One has a horizontal line across the page with looping lines representing time travel above and below it showing leaps forward and backward. It’s clever but it doesn’t address the new timelines spinning off of the original with each decision in the past rippling forward.
The other method is a series of horizontal lines across the page, each representing newly formed timelines by character action taken and diagonal lines connecting them representing the forward/backward travel. This method is possibly more accurate but also clunky and “heavy” (as Marty would say).
I wanted to take an approach that took the accuracy of one and the elegance of the other. Maybe it works, maybe it doesn’t, but I’m all about trying and failing in public.
It’s how you grow.
AM: I understand you’re a member of the Feedback Loop project. Please explain a little what that is, what feedback did you get?
ZB: The Feedback Loop is a more formalized version of what I do every day. Whether it’s consulting with my colleague David Kelly (@DataDavidDeluxe) or reaching out to twitter friends like Kate Schaub (@schaubkatelyn) or Lindsay Betzendahl (@ZenDollData) for feedback on a personal project, I’m always asking for feedback and giving feedback.
The Feedback Loop is an exercise in that for a group of about 20 people who cycle through in various “classes”. We have a Slack channel, and users agree to be respectful and supportive all while helping the other participants hone a particular project.
In my case rather than putting out a new project I am attempting to resurrect a failed viz of mine about San Francisco municipal employee salaries. When digging into the dataset I found 4 massive outliers in terms of salary that took over my total focus of the project.
Who are these people? Why are they making SO much more than everyone else (and even in some cases others with the same title). Why are these flies in the ointment wrecking my dataset?
I did a ton of news searches online, digging out the stories of who they were and why they made so much money in a single year, but ultimate the viz went off the rails and remained unfinished.
I needed help.
AM: How did you decide to apply the feedback?
It’s a work in progress. I’ve learned a lot from digging through the data that I can apply to my future state. Originally I had this whole “Flies in the Ointment” approach where I was calling out the 4 major outliers that were distracting me from actually digging into the salary data. But that slowly became the entire viz, and that’s not a super fruitful take.
I’m doing a rework, almost from the ground up. The best feedback I’ve gotten is that the theme is a distraction from the data. As a guy that likes to get creative and try to make things fun, that’s something I definitely needed to hear in this case. I went way overboard with pictures that don’t really have anything to do with the data and analogies that distract from reality.
I’ve done some “data journalism” type work in the past, like on my Pakistan Election Year viz about the political violence and assassination surrounding the 2018 election. In that case I actually removed color as an option for myself and just dug. I read articles and news stories, and I did my best to show the who and where and why of the real-life political violence. I actually went a little too far with my self-imposed “No Color!” mandate, and at Kevin Flerlage’s (@FlerlageKev) suggestion added a little bit back in.
I think this can be something similar, I just need to adjust, refocus, and address my outliers and just move on.
AM: You received a ton of Tableau Public Viz of the Day comment calls on Twitter and so much ‘love’ sent your way. Were you surprised it registered on that scale (especially without animations or newfangled bells and whistles)?
ZB: I’m always surprised if anyone likes something I do (haha). If you look at my public portfolio you can see that I definitely have a “lane”. I like to have fun, look into things that I’m personally into whether it be comic books, film, video games, etc. Many of these topics aren’t for everyone, but I try to approach them in a way that I find interesting and hopefully even someone who’s not into that topic might as well.
It’s like a good documentary. I can’t tell you how many I’ve watched (or podcasts I’ve listened to) on topics that I don’t consider myself interested in. But with the right approach and in the hands of a storyteller who is passionate, that topic can capture the imagination of nearly anyone.
It’s always fun to see people respond to something that might not be in their “lane”. I have no interest in sports, but have seen some sports vizzes that drew me in and just had me digging to learn more (Your/Brian Moore’s (@BMooreWasTaken) Sandberg Game viz is a great example). (1)
In the case of my McFly Timeline viz…like you yourself said when I was asking for feedback, it’s controversial, and that is one of the reasons the positive response was such a surprise for me.
By my estimation, there were two main faults someone could find with it:
1. The chart type
2. The color choice
The chart type is…novel. Expressing time when time isn’t a straight line is an interesting experiment. I reached out to several friends and for the most part people were on board with it, including Steve Wexler (@DataRevelations); which was a shock.
The color has been the more controversial choice ultimately, people love it or hate it. It’s being called “Red” but this describes it as “Sunset Orange”. I sort of see it as a deep salmon. Color is surprisingly subjective, both based on our individual perception and the tuning of the screen it’s viewed on.
When I first mocked this viz up, it was a white background and a faint tan line. Once I knew where I wanted the viz to go, I realized I wanted to punch up the “HEY, LOOK” without adding a ton of clutter and elements. The color was a good way to get those eyes without compromising the content.
I wouldn’t use this for a business viz, the color is definitely an “alarm!” color that wouldn’t be conducive to measured decision making, but it was perfect for getting eyes, then converting those eyes to attention.
“My Special Day” Viz
Music has always been a substantial part of my life. Since emotional depth is a restriction of mine, songs and melody helped me understand and learn how people felt and dealt with ‘human’ situations and gave me something I could use to connect with others. It also was something that helped introduce me to the nerdy fun world of data analytics (even though I had no idea what it was at the time).
Between ages 9–13, I always listened to (and often recorded on cassette tape) Casey Kasem’s weekly Top-40 countdown and compared it to the Top-20 of Billboard’s Top-200 song countdown in the Sunday newspaper. I kept records comparing discrepancies and attempted to forecast positions and likely peaks based on a song’s progression — especially if it was from one of my favorite artists. I didn’t share this hobby with others — it was ‘my’ thing. When I grew into my teens and identifying with my Gen-X peers, pop music went to the wayside and searching for ‘rare’ music gems discoverable by word of mouth, indie record stores or college radio. After a decade plus of avoiding saccharine swoons, bursts of nostalgia came pouring through.
At Barnes and Noble, I jumped at Joel Whitburn’s Book of Top-40 hits, so I could rediscover the songs I missed or forgotten as a kid. It became a new hobby in my late 20s to try to find and listen to all the songs on a week from artists with Top-40 hits. Since then this desire to reach back hits me in waves and something amazing happened…
In 2018, I found Sean Miller’s (@HipsterVizNinja) Billboard dataset covering Billboard Top-100 hits from 1958–2017 — it was the one set that started my interest in creating vizzes for fun rather than just work dashboards. My goal at the time was to create a data visualization/database for myself and maybe nerds like me. It was enjoyable, so I created a couple more (clunky) vizzes from it.
On 2/19/20, a new bolt of inspiration hit after remembering Sean updated the dataset through 2019 and added a separate Spotify attribute database for the charted songs. I knew I could make a weekly Top-100 list from August 1958 through 2019 with Spotify attributes. 100 songs are just too much/overwhelming for most, so I need to shrink the list (more on that in a bit)… then I remembered Zach Bowder’s ‘Happy Birthday’ viz and Brian Moore’s engaging ‘DataViz Name Tag’ . How about I develop a weekly Top-10 list based on your birthday with customizable designs?
Unlike my music vizzes of the past, this viz is less analytical and just fun. How can I level-up the fun-factor even more?
Learning from the vizzes referenced from others, I knew it should totally have the capability to interact with the music. To make it bite-sized enough, I decided to grab only the Top-10 lists for each week. Initially, I went with week to week, but it became a filter mess when weeks overlapped with prior months (i.e. if you were born on the 1st or 2nd, you would have to change your filter) — I thank Kevin for finding that out. To fix that problem, I densified the data to daily. The chart had a Saturday beginning, so I made a few calculations in Excel to fill Sunday-Friday, so every day was selectable (if the day selected was between 8/2/58 through 12/31/19). Gap problem solved.
Making the viz customizable was the next piece. Pretty early in the viz incubator (my brain), I decided to make various stacked shape/scatterplot chart of the Spotify attributes — these attributes don’t have a ton of value to the casual music fan (values include variance, danceability, instrumentation and etc. from 0–1), but can be used for each song in the Top-10 to design a stacked shape chart. The design choices before was Tableau icon, Pac-Man, flames icon from PowerPoint + unfilled circles and dots — these were okay, but with tinkering, I decided to keep the Tableau icon and create icon .pngs from PowerPoint adding stars, hearts and pies (chart) while booting Pac-Man, circles and squares. In addition, I wanted a ‘colorful’ option (different color per song) and a ‘mono’ option (single color for shapes per year of song charting). With five shapes and colorful/mono options for each, a user has 10 options for designing their music card.
Other design functions…
- Creating hidden containers to see the full design was something that made the viz more of a ‘card’ feel. A user can hide the instructions, filters and Spotify player to be able to snag a photo of their custom card.
- A YouTube song search for each song on the list — I did a YouTube search, found the search link base of a YouTube search and tested adding plain text following the link and it worked as intended. This allowed me to add a URL action with the base search address and add the ‘performer — song’ field.
- Adding a simple ‘type-in’ parameter — The initial viz intended to be a ‘birthday’ viz but realized some people who used the viz may have been born before the 8/2/58 date. With the type-in parameter, I can reference it in the worksheet title to change the reference of the card. Note: It defaults to ‘My Special Day’ card. Note: I forgot to change the ‘Happy Birthday’ image on the bottom of the viz until last week when I made a ‘My Special Day’ graphic :face_palm:.
Between 2/19–2/25, I held my ad-hoc feedback loop sending it to Sarah Bartlett (@sarahlovesdata), Sean Miller, Zach Bowders, Brian Moore, Toan Hoang (@thoang1000) and Kevin. They all had great feedback and encouragement.
On 2/26, it was released and a source of a lot of fun lists and discussion in the community. It didn’t win any special recognition, but it came out exactly as I wanted it to and had a blast finding more inspiration with Sean Miller’s great data grab. Zach Bowders was inspired to put his take on my viz it and made a fun little viz relating what music your parents were listening to when you were made. As Tableau Public publisher who views viz-making on the site as a fun hobby, crafting in your viz-ion and sharing with friends is rewarding — inspiring others (even if it is silly takes on your work) adds to the delight.
My mind wasn’t great this week. I’ve been in a bit of an autistic depression/anxiety funk with a feeling of a lack of purpose. Fortunately, I recognized it in the middle of the week and discovered why on Friday and am dealing with it. This song basically mirrors my vibe of that period.
Very good, but not great documentary. The editing is ADHD meets Grindhouse and missing features on some major performers (Jet Li, for example), but I enjoyed it on Netflix for the abridged piece of cinema history it was.
1) Aww, thanks!
2) Weekly health metrics (through 7 weeks):