40years of Ebola: Visualizing Ebola virus historical timeline
Role: Designer & Data Analyst
Team: 1 Designer, 1 Developer (Apurva Nagarajan)
Project Duration: 4 weeks (December 2018)
Project Overview
As part of our data visualization semester project, I designed the Ebola virus’s historical timeline, exhibited at the Purdue University’s Data Visualization Exhibition. The visualization was later selected as one of two works created as a visual analytics tool in collaboration with the Computer Graphics Technology department. The tool provides a historical timeline of all countries impacted by the EVD from 1976 to 2018. We show the type, severity, and spread of EVD that affect countries throughout the years. We developed this tool to help the general public to understand more about the Ebola virus.
My Role (Individual)
→Problem framing
→Data sourcing and analysis
→ Multiple rounds of ideation and sketching
→ Visual encoding and design
→Testing, evaluation, and Design QAs
→ Video for storytelling
→Design handoff through zeplin
→HTML Markups
→ Maintaining Git collaboration on Github.
My Role (with Apurva)
→HTML Markups & D3.js
→ Maintaining Git collaboration on Github.
Background
The Ebola Virus Disease (EVD) is a rare viral disease transmitted to humans through contact or bodily fluids. It was originally observed in then Zaire, now, the Democratic Republic of Congo, in 1976. The virus is believed to spread through bats and bush meat. In 2014, there was an outbreak in several countries in West Africa, which killed tens of thousands of people. Since August of 2018, there has been a new EVD outbreak in Congo, which has drawn worldwide attention from medical professionals to governments at various levels. As a result, I decided to investigate the virus’s historical pattern and the impact it has had on the various countries it has afflicted.
Understanding the virus
The Ebola virus is one of the deadliest viral diseases. EVD most commonly affects people and nonhuman primates (such as monkeys, gorillas, and chimpanzees). The genus Ebolavirus has several strains, including:
- Sudan virus (species Sudan ebolavirus)
- Taï Forest virus (species Taï Forest ebolavirus, formerly Côte d’Ivoire ebolavirus)
- Bundibugyo virus (species Bundibugyo ebolavirus)
- Reston virus (species Reston ebolavirus)
- Bombali virus (species Bombali ebolavirus)
Research
I conducted secondary research to understand the virus, existing data visualization, and how sensitive medical datasets are visualized. Following that, I established my data sources from the Center for Disease Control [CDC]. The data was directly derived from the CDC website on the Ebola virus. The website is constantly updated, ensuring that the dataset is up to date. After determining the source, the data was collected and analyzed to determine the key details that could communicate the context of the outbreaks to viewers.
Analyzing and defining the data
The data consisted of the key identifiers like year of the outbreak, reported number of cases, reported deaths, survival rate, country impacted, and Ebola species. With all of this data being temporal, the most relevant information to convey was a historical timeline of the outbreak’s years. The minor details to be communicated were viral attributes, which described the element. The location was relevant in showing the country of the outbreak.
Viral attributes on a timeline must be linked to the location and depict death and survival rate.
Sketching & Ideation
During the ideation and brainstorming stage, I explored several options for visualization timelines, health, and disease, etc. With the general public as the audience, my goal was to present a visualization that was easy to understand. After several ideations, I took inspiration from the Ebola virus structure to tell its history.
With the glycoprotein structure being the most obvious part of the virus, I concluded using it to depict the deaths, survival, and cases. During my research, I also found that most people before and even don’t know that Ebola had been identified in other countries outside of Africa. This also made me consider highlight the locations as well.
Visual encoding and Design
One of the most difficult parts of the design was the color encoding and shapes' choice to depict the virus. The initial concept was to use a bubble (mushroom) to represent the cases, death, and survival, but I realized it would occlude the map upon implementation. I reconsidered the shape, which led me to use a more geometric form that looked visually pleasing.
Defining the Legend
A considerable amount of time was spent to refine the design components and the legend particularly. With clarity and simplicity as the goal, I had to trade stylistic elements in favor of the dots, representing the survival and the death rate.
Final Interactive Visualization
After the final design, my work was selected as one of two designs to be developed in collaboration with the Computer Graphics Technology Data Visualization department. I was matched with a developer — Apurva, for us to build the functioning prototype. Over three weeks, we collaborated on Zeplin, Github, and email.
User Testing & Evaluation
Once the functioning interactive prototype was developed, it was time to see how it performed with people. I invited willing participants to go through the user testing and evaluation and get their opinion of the tool. Users were to give their feedback on the five(5) high-level metrics.
7
Participants
Most of these participants didn’t have much experience with data visualization; the rest were colleague interaction designers.
5
Metrics Measured
The success metrics were Perceptual, Cognitive, Usability, Tasks, System.
The process & Results
Each participant was presented with the visualization to explore and evaluate it based on multivariate heuristics from Tufte, Schneiderman, Norman Heuristics. Participants were encouraged to think aloud as they walked through the system. Since most of the participants were new to the think-aloud session, I had to train them to articulate what was going through their minds. Each participant completed an exit questionnaire, which they had to rate. The results from all the participants were aggregated into a single 1–5 scale ratings, as seen below.
Key findings from User testing
Overall, users found the visualization to be accessible and aesthetically pleasing. Users found it readable and legible. Users found it easy to comprehend the content and the over visualization. The legend was effective and was the first point of interest for most users. While most users found the filtering systems useful, their toggle system was had to use or remember. Contrary to what I thought was minimalist most users found it to be somewhat cluttered.
Conclusion
Reflection
This project taught me a lot about developing and conducting usability testing on data visualization projects. Through the testings, I learned how to consider the domain or audience of the users before encoding the visualization. I also got to experiment with D3.js and basic javascript. I discovered that all design heuristics converge in some way to eventually benefit the user.
Other discussions
A few topics that were not fully addressed in this case study that you might contact me about in person:
- Prototyping
- Design iterations
- Choice of design Interaction
- Initial design concepts
