Various Projects

Whiskey: An International Multivariate Visualization

Information Visualization, September 2014

The Problem

Given a very large set of multivariate data, the assignment was to include as much of said data as possible in a single, one-page static visualization.

Looking at the data, I came up with the following questions:
• Which country produces the most of different categories of whiskeys? Overall number of whiskeys?
• Of those whiskeys, which are the most highly rated by country and category?
• Is the price justified by rating? That is, why buy the $3,000 Scottish Highlands bottle if you could get the same variety at the same or slightly higher rating for $55?

The Visualization

This Gantt chart, created with the information visualization toolkit Tableau, examines whiskey prices (as marks) by ratings (divergent color scale). The whiskey ratings are broken down by country (e.g., Scotland; U.S.), and then further by category (e.g., Bourbon; Blended). The length of the bars indicates the count of how many whiskeys in each category. The ratings are shown by colors, which go from dark to light green (very good to good), yellow (average/below average), and orange-red (poor). While not ideal for color-blind users, this presents a familiar “traffic light” design for users to quickly get an idea for whether something is highly or poorly rated.

The chart shows different characteristics of the data at different levels.
• Looking at the length of the bars, we can see which country produces the most whiskeys of given categories, and by looking at all of the bars for a country, one can get a general idea of how many whiskeys overall are produced.
• Looking at color, we can see broadly which whiskeys are highly rated by category and by country, and how this applies to the amount of whiskey produced (e.g., Canada produces many blended whiskeys, but they are mostly average to “bottom of the barrel”).
• If we look more closely, we see actual prices for each whiskey.

I based the ratings scheme on the scale used by Whisky Advocate to determine the colors for ratings. Below is the raw static visualization exported from Tableau (click for a larger version).

Initial Tableau visualization

To finalize the print version, I used Illustrator to add a key based on the rating bins I had coded rather than the spectrum provided by Tableau, and cleaned up some text that was not displaying properly (click for a larger version).

Edited Tableau visualization


Tableau, while very powerful and capable, can be tricky to navigate for finer details, like creating custom color palettes for ranges. I wanted to have my categories/ranges match up exactly to Whisky Advocate, but issues with the custom ColorBrewer color palettes from the Preferences file and issues with binning ranges of numbers prevented me from doing so in this iteration. The actual, dynamic Tableau visualization is more robust because the user can mouse over individual items to see brand, price, age, and rating.

Lessons learned & related work

I continued working with Tableau later in the course as well as other information visualization toolkits throughout the semester. The class taught me a great deal about the importance of telling a story with your data. Like a good story, a good visualization is memorable, but not overembellished; it shows rather than tells. And, as William Shakespeare himself once said, “Three-dimensional pie charts are nobody’s friend.”*

While this and all other short assignments were individual efforts, the most fascinating project in this class was a team effort. If you’re a cinephile, be sure to check out our team’s original visualization, coded in D3, for building customized teams of cast and crew: Making Movies Matter.

*It is highly unlikely that Mr. Shakespeare ever actually commented on pie charts, 3D or otherwise.

Buddy the Seatbelt Bear

Personal development, March 2014

The Problem

Working with Applyosaurus rekindled an interest in sewing and soft materials. Modifying an existing plush/stuffed animal was very interesting, so naturally I wanted to make my own from scratch. Opportunity knocked in the form of spring break.

I’d been noticing some soreness in my left shoulder area while driving, likely in part due to the cross-body seatbelt cutting into the engaged area. I wanted to make something that would be useful in daily life, and decided on a stuffed bear that doubled as a seatbelt cushion to relieve pressure on the shoulder and chest while driving.

Design goals were to create a product that looked like a stuffed animal but was soft and effective as a cushion, durable, adjustable for multiple seatbelt heights and positions, unobtrusive so as to not obstruct movement or vision, and, of course, adorable.


As someone without much sewing experience, I started by researching stuffed animal construction and, more specifically, seatbelt cushions. Some existing products, like commercially available large pillow/seatbelt attachment animals for children, weren’t quite appropriate for a driver, while traditional just-the-facts-ma’am cushions were pretty bland. A valuable resource for adapting these designs was a blogger who created her own animal cushions for her children and documented the construction in detail.

With these ideas for inspiration, I went my own way. I bought some corduroy-like fabric, polyfill stuffing, and colored thread, and it was off to the races.


Buddy the bear is entirely hand-stitched. I eyed the dimensions of my seatbelt and the appropriate coverage area to cushion my shoulder, then sketched a simple pattern. From there, I sewed the head, torso, and limbs individually, then stuffed them with polyfill and sewed them to the torso.

Limb with polyfillSewn torso and head
I braided scrap threads to make button-loops, then sewed buttons and button-loops to the arms and legs so Buddy would stay in place on the seatbelt.

Unbuttoned arms and legsButtoned arms and legs

Finally, I cut and sewed the ears, then ‘embroidered’ the face with a double-threaded needle.

IMG_20140320_151841'Embroidered' face

Finished product

Buddy on belt

Success! Buddy fits perfectly on the seatbelt and stays in place while driving while sliding up and down easily if adjustments are needed.

With Buddy as my co-pilot, the extra cushioning made a difference — driving with Buddy is more comfortable, not to mention more cheerful.

Posters for Interactivity: 2015 vs. 2014

Georgia Tech MS-HCI Program Event, February 2015 and February 2014

Georgia Tech’s MS-HCI and MS-Digital Media students present their work to potential employers and other interested parties annually at Interactivity, a ‘reverse job fair.’

These posters were both made with InDesign. I think it’s a lot of fun to see how things have progressed in the span of a year. Click the posters for larger versions!


My design aesthetic has developed considerably since last year, which is always a good sign. I concentrated on creating a clean, color-consistent layout with less text and plenty of images.Since I stay with my poster nearly the entire time, I can walk viewers through the projects in an image-heavy poster without the need for much text. Many printed posters later, I’ve also learned that the poster printer adds extra margins, so (a) it’s good to use a white background to prevent unintended contrast, and (b) there’s no need to leave a lot of margin space since the print will look somewhat different.

Click here for a PDF version.

2015 Interactivity poster


For a fun exercise, I mocked up the Adobe CS6 and Microsoft Office 2013 icons used here in Illustrator. No tracing required, although I did use reference swatches from the internet to color the CS6 icons). On that note, no copyright infringement intended!

Elena's 2014 Interactivity poster