Visualisation of Average and Median Dengue infection rates across countries


#1

I’m a student from Swansea University looking to visualise some data from the Tycho project - namely that on the Dengue virus across all countries for which there is data. Visualisation made using Tableau.
Countries: Brazil, Vietnam, China, Thailand, Cuba, India, etc.
Disease: Dengue virus
Years: 1955-2009
X-axis is mapped to average number of recorded Dengue cases for each country across all years, y-axis mapped to median across all years. Each different colour represents a different country - in retrospect, this is fairly redundant, only serving to make multiple clustered points more distinguishable from each other.


My intention was to easily identify countries where the mean value, which is easily skewed by outliers, and the median value were either very close or very disparate - the former implying Dengue cases to remain relatively stable, the latter implying a tendency toward sudden outbreaks.

My questions are as follows:
Should I have opted for utilising logarithmic scales on the axes in order to make the differences between tightly-clustered datapoints more visible, or would the nature of such impede at-a-glance understanding of the data?
Moreover, would another form of visualisation be better suited to showing the relationship between average number of Dengue cases and the median?


#2

Hi,

I might not understand exactly what you are trying to do as it seems that you are trying to show the relationship between Median and Average and I believe that they always are related mathematically although I might be incorrect.
Wouldn’t it make more sense to show the median or average in relation to each country? or each year for one country? Just throwing some ideas at you.

With regards to the scale I would just try it out and see if it serves your purposes better.

Best of luck

Carlo


#3

I’m a bit baffled that you have different scales for the two axes: 0-90k vs. 0-8k. Are the differences that great?

To your question, yes, I’d use a logarithmic scale, add country names, and a tooltip to show actual values.

Alternatively you could also just use a country list and show both average and median as bars. Order the list by their ratio.

Or, have a list of countries and for each show a small bar chart with data over years might be better at showing trends and outliers (a heatmap where rows are countries, columns are years, and cells are counts might also do)