Visualising cases of Measles, Influenza, Typhoid, Tuberculosis and Pertussis in the US

Visual Design
#1

Hi,

I am a student at Swansea University who has been asked to visualize some data from the Project Tycho dataset (https://www.tycho.pitt.edu/).

I have created a Radar Chart to visualize the difference in infection numbers of 5 different diseases in the US. Here is a link if you wish to take a closer look: https://plot.ly/~johnmctiernan/24/new-york-california-virginia-texas-illinois/#plot

• Image:

plot%20from%20API%20(12)
plot%20from%20API%20(12)1443×1024 145 KB
!
• Visual Design Type: Radar Chart
• Name of Tool: Plotly
• Country: United States
• Disease: Measles, Influenza, Typhoid, Tuberculosis and Pertussis
• Year: 1888-2017
• Visual Mappings: Each colour represents a different state as shown in the legend. The diseases each have their own axis that visualises the count value of that disease for a given state. Each ring/circle represent 0.5 million people. We are also able to remove any of the states, if we wish to compare a smaller number.
• Unique Observation: We can see the Measles is the most infectious disease out of the five shown, as it consistently has the highest count value for each state. We can also see that even though the population of California is higher than Texas, the spread of measles in Texas over the past 100+ years has been much greater than that of California.
• Data Preparation: The sum of the count values had to be calculated using python syntax.
• DOI: 10.25337/T7/ptycho.v2.0/US.6142004
•Literature: T. Munzner, Visualization Analysis & Design, A. K. Peters, 2014. I used this literature when deciding if I should use this visualisation. Specifically, I looked at the use of length and area as a channel for visualisation. From this literature, I could see that position on a common scale, as well as area and length are good channels for the visualisation of ordered data.
My questions are:

  • Is there any better way I could represent this data?
  • Would it be a worthwhile time investment to add a time slider to shift through the years to compare yearly numbers?
#2

Hi,

Before I answer your questions I’d just like to say that this is a cool design, it seems very unique compared to some other designs I have seen and it seems like a great way of showing two categories (States and Diseases) in one graph.

As for question 1, although I like the design one issue I can see is that the states all overlap, which could make it difficult for some people to analyse, I think either a simpler stacked bar chart design or even a small multiples design with a single radar for each state would be good alternatives as both of these would prevent overlapping

For question 2, it would definitely be interesting to have a time slider to perhaps filter view more recent disease distributions, since your time period is so large (1888 - 2017) I would probably recommend a slider based on decades (Also this would make the design more closely follow
a part of Ben Schneiderman’s mantra, “zoom and filter”)

#3

In general, a radar chart is a great tool to be used for comparisons. However, in this case, the infection values for the states are quite similar, and therefore the polygons have a similar shape. In addition, the infection rate for Typhoid is always 0 for all states. So in this case I would rather use a bar chart, or a stacked bar chart.