"Dear Data" Design challenge
As part of the interview process for a UX Design position, the design team at Quid challenged me to create a data visualization inspired by the Dear Data project using a data set containing milage traveled in one week. Below I describe my process for creating that data visualization.
SNEAK PEEK OF THE FINAL PRODUCT ^^
Here's an angled, partially obscured photo of the final piece. In this shot I was going for "I'm a designer and I make important things, but I also went to art school and took a photography class so I would know how to frame shots of sketches on paper for my portfolio." Read on for the good stuff.
STEP 1: CHECK OUT DATA + ASK SOME QUESTIONS
I started by checking out the dataset, scanning quickly for key words, patterns, and anomalies. For example, I noticed that the average pace and duration weren’t available for each entry, so I had to think about (a) what value including these categories might add and (b) how I could / would normalize them if I decided to include them.
Based on the columns, I thought about what questions would be interesting to ask of the data—what story could it tell? For a moment I pretended the data set was my own. I thought about what I might be interested in knowing if this was data about my own week, and I wrote down some questions that I wanted to ask of the data.
STEP 2: START SKETCHING
Then I just dove right in, using pencil and paper to see what would come out. Initially I took a really simple approach: I just wanted to know the answer to “where did I go this week?” I started sketching focusing only on the destinations, the order in which they were visited, and the days of the week.
I got partway through and was bored by the story that was being told—it wasn’t interesting enough yet. I needed to add more layers of data to get to the real juicy insights later (*ominous music*). I took a pause and noodled for a while about how I could keep the visualization simple while layering additional data types.
STEP 3: SHAPES, DOTS, LINES
I went back to asking some questions:
- How far did I travel each day? In comparison to the other days that week?
- Who was I with?
- How’d I get there?
I mapped the column headings from the spreadsheet to the questions to determine what combination of inputs I’d need to answer these questions.
Then I started creating iconography to represent each data category—shapes for destinations, dots for people, colors for destination types, lines for mode of transport.
STEP 4: MORE SKETCHING
Then back to sketching. I started by measuring out the milage. Thursday was the longest milage traveled, so I did that one first to make sure I had enough room on the page. Perfect fit! Almost.
STEP 5: PAUSE, CONSIDER, SIMPLIFY, CONTINUE
After all the destination placeholder boxes were plotted, I realized I could simplify the iconography by only dividing boxes that contained 2 destinations in the same trip (this happened once on Thursday and once on Friday)—much better.
STEP 6: WHERE DID I GO?
Instead of giving each specific destination its own color, I grouped them by category: home, work, food (lunch, dinner, coffee), kids (school, park), and errands (post office, errands). I coded each of the destination shapes with the appropriate color.
STEP 7: WHO WAS I WITH?
Then I added the people dots. Now it’s starting to get interesting!
STEP 8: HOW DID I GET THERE?
Finally I drew the “transportation type” lines. I was pleased with this layer especially because it was something I had written off initially as non-interesting, but it ended up really rounding out the visualization. Yay for iteration.
STEP 9: DATA VIZ DECODER
I added the key for reading to the visualization itself in the spirit of the Dear Data project—I love how each visualization is a perfectly imperfect, self-contained analog unit.
Check it! A Dear Data-inspired visualization in all its analog glory. (Can you tell I really enjoyed this project? And, yes, I joined the Dear Data Google Group to find a data viz pen pal of my own...)
WHY DID YOU CHOOSE TO VISUALIZE IT IN THE WAY THAT YOU CHOSE?
I chose a very low-fi, analog output format for a few reasons:
- To channel inspiration from Dear Data. One of my favorite parts of the Dear Data project was its analog nature. It seemed like doing so many visualizations really forced the authors to get creative with their visual languages. Additionally, I appreciate that each was a self-contained unit—the visualization and the “how to read” tips were all contained to one postcard. I used each of these elements to shape my own final product.
- To increase experimentation and iteration. In my experience, when you view something as “precious,” you’re less likely to take risks with it. When it comes to balancing your Grandma’s heirloom china on your head that’s probably okay, but when it comes to design, trying to bring something to visual perfection too soon in the process can stifle creativity in the ultimate output. I just feel so much worse scrapping work if I spent a bunch of time making sure all the lines were evenly distributed. During this project I realized after I had already drawn some black marker lines that I wanted to change one of the icons, so I just folded the top of the paper over and re-drew that day’s worth of data—the distribution of the lines was a little wonky, but it didn’t really matter. Using pencil and paper makes the stakes seem a lot lower.
- To add constraints. A design is nothing without constraints. If there are no explicit constraints at the outset of a project, your design will probably fall prey to assumed-but-unexpressed constraints from stakeholders. I’ve worked on projects with “no timeline” that were almost abandoned because stakeholders were disappointed that things were “taking too long.” It’s better to set some constraints for yourself and re-consider them later than to toil away thinking there are none. For this project, I set the constraints of “2-4 hours of work, maximum” and “analog, hand-drawn, self-contained unit for the output” as constraints for myself (based on your instructions). I had so many big ideas; I could have spun my wheels indefinitely if I didn’t set constraints early.
WHAT PROPERTIES OF THE DATA DID YOU CHOOSE TO REPRESENT WITH YOUR VISUALIZATION? WHICH DID YOU CHOOSE TO NOT REPRESENT?
Chose not to visualize:
- Average pace
Chose to visualize:
- Date (partially; I only used the day of the week)
- Time (partially; I used chronology as a proxy for time)
- Start location
- People in car
- Car type
WHAT ARE THE IMPORTANT INSIGHTS THAT VIEWERS SHOULD BE AWARE OF?
- The first couple stops of the day are pretty routine-oriented: the subject starts the day by taking her daughter to school and then on to work (a run crept in there one day).
- It seems that the subject and her spouse either work together or in very close physical proximity to one another. They don’t go to work together in the morning—I would hypothesize that the spouse takes their son to school or daycare and then they meet at work / arrive at work separately. Additionally, the subject and her spouse often travel to lunch together or spend time together at home around mid-day and then travel back to work together later.
- The subject went out for food every day except Friday.
- Total milage traveled on Monday and Friday were much less than the mid-week days, which were all relatively comparable to each other. (I would be interested to see if this persists across multiple weeks and how it may change in relation to the milage traveled on the weekends)
- It looks like the kids are typically picked up by either the subject’s spouse or another unknown party since daughter and son are mostly absent from later-in-the-day travel (although a pick-up of the subject’s daughter did occur on Friday). Spouse, son, and daughter were also present for family dinner out on Thursday.