Visualising Pneumonia vs Influenza in the US using a treemap

Visual Design
#1

Hi,

I am a Computer Science student currently studying Data Visualisation and as part of my coursework I am exploring and visualising the Project Tycho dataset. Specifically, I’m comparing the number of infections and fatalities of Pneumonia and Influenza in the US from 1920 – 1929 (DOIs: 10.25337/T7/ptycho.v2.0/US.6142004 and 10.25337/T7/ptycho.v2.0/US.233604007). https://www.tycho.pitt.edu/

I have visualised this data using a treemap, with a hierarchy of State -> Year -> Disease. The size of the rectangles represents the number of infections and the colour represents the number of fatalities. I created this visualisation using Tableau and although it is interactive so all the data can be seen through the interactions, hovering the mouse over a rectangle will show all the data for that rectangle.

I have tried to stick to Tuft’s principles [1] and so I was wondering if there was a modification I could make, to the treemap or data, for me to more clearly show all of the data for all of the states? As the ones with fewer infections are harder to see. Also, are there any other improvements I could make to the visualisation.

[1] Edward, T. (2001). The visual display of quantitative information. Graphics Press, Cheshire, USA.

pnemonia%20and%20influenza
pnemonia%20and%20influenza1920×1080 280 KB

#2

Hi @MPayne,

I’m not familiar with Tableau, but is there a way to have the size mapped non-linearly? Doing this would make the smaller rectangles, larger while still keeping the integrity of the visualisation.

#3

Hi, MPayne,

  1. You may experiment with a different ordering, such as, Year -> State -> Disease, Year -> Disease -> State, …, and try to write down what can you observe. You may compare the usefulness of different orderings.

  2. Please note that colour perception is affected by the size of the rectangles (especially when they are small, e.g., bottom right of the image). See SZAFIR D. A.: Modeling color difference for visualization design. IEEE TVCG 24, 1 (Jan 2018), 392–401. You may use a multi-band colormap to alleviate this issue. see BORGO R., et al. Evaluating the impact of task demands and block resolution on the effectiveness of pixel-based visualization. IEEE TVCG 16, 6 (2010), 963–972 and WARE C.: Color sequences for univariate maps: Theory, experiments, and principles. IEEE Computer Graphics and Applications 8, 5 (1988), 41–49.