Visualizations that Save Lives.

"Remember John Snow!"  I shout like a lunatic at a classroom full of analysts as they dutifully file out of their latest BI training.  No, not Jon Snow of the Night's Watch (Game of Thrones fans unite!) but John Snow M.D. of London, in 1854.

Sometimes as analysts we lose the forest for the trees.  Take a moment, and think about times where operating at the macro level would have pointed out a whole different line of thought.  Too often our smartest peers forget this, and need to be reminded that quality (with very few exceptions) defeats quantity every time. 

John Snow was facing a cholera epidemic in London that was killing people faster than they could be buried.  Dr. Snow had the foresight to create a visualization that stacked all 578 deaths by home address, and he noted a concentration near the Broad Street water pump.  This outlier gave him everything he needed to make an educated hypothesis about the source of the outbreak.  

When we are looking at our data, it's important to resist the urge to look at the very line level detail, and fight for the outliers.  A creative visualization can lead the user down breadcrumb paths of inquiry that they might not have followed.  This combats the common situation where an analyst starts with the end in mind.  They end up using the data to prove a hypothesis, rather than let the data lead them to the answer.  

Sometimes people complain that visualizations take up too much space, or show too little - but that is not always the case.  Take for instance the Marimekko chart:

Every pixel of this visualization is telling you something!  The x and y axis tell you about the intersection of the data, the width of the columns and rows show you the relative aggregate weight.  The chart above is the equivalent of 10 pie charts, countless data points, and comes with the added benefit of relative scale.  You can easily see that Category 1 (the left most column) has significantly more weight than column 7.  If those were two pie charts, you could easily see the difference between blue and brown, but would have no idea that 1 was potentially more significant than 7.  The ability to understand these relationships by looking at the line detail would be nearly impossible.

Being a good analyst or data scientist (a much abused title) is as much art as it is science.  A clever use of visualization is the most powerful opportunity we have to bring art into the science of information.  If John Snow used a viz to save lives in 1854, it's possible that you can do the same today.  

Remember John Snow!