Michel Baudin‘s comments: On two prior occasions, I pointed out remarkable graphics in the New York Times:
This one, however, is remarkably rich, yet readable thanks to details like labeling the lines with the name of the country, as opposed to having the names in a legend block. The author, Josh Katz. is a graphics editor for the newspaper and his co-author, Margot Sanger-Katz, a writer on health care.
On the newspaper’s website, if you hover over one of the country lines, the others dim out and additional details pop up, including a trend line, as in, for example:
Click picture to enlarge
The value of the logarithmic scale on the y-axis is that exponential growth shows up as a straight line. When exploring relationships between variables, it’s a classic engineer’s trick to scatterplot them first in their raw coordinates, then with one axis logarithmic and finally with both axes logarithmic. If it looks straight in the first plot, you try a linear model; in one of the semi-logarithmic plots, an exponential model; in the bi-logarithmic plot, a power model.
On this chart, the US curve is frighteningly straight, with the number of deaths doubling in slightly less than 3 days. China’s curve is concave, meaning that it is no longer growing exponentially. The word “exponential” is loosely applied in the media to any kind of rapid growth but it has a precise, technical meaning that is honored on this chart: the next increment is proportional to the current, cumulative value, like compounding interest, and it’s why you can describe it in terms of the fixed time it takes for the value to double.
We pay attention to this chart because the information in it is of vital importance to us. Its design, however, is essential to our ability to understand it. It’s a long way from the pies and stacked bars that are the mainstay of factory performance boards.
Mar 24 2020
A Sobering But Remarkable Chart | Josh Katz and Margot Sanger-Katz | New York Times
Click picture to enlarge as needed.
“As the coronavirus pandemic unfolds, people are dying around the world. But the trajectories of cases and deaths differ by country.”
Source: The New York Times
Michel Baudin‘s comments: On two prior occasions, I pointed out remarkable graphics in the New York Times:
This one, however, is remarkably rich, yet readable thanks to details like labeling the lines with the name of the country, as opposed to having the names in a legend block. The author, Josh Katz. is a graphics editor for the newspaper and his co-author, Margot Sanger-Katz, a writer on health care.
On the newspaper’s website, if you hover over one of the country lines, the others dim out and additional details pop up, including a trend line, as in, for example:
Click picture to enlarge
The value of the logarithmic scale on the y-axis is that exponential growth shows up as a straight line. When exploring relationships between variables, it’s a classic engineer’s trick to scatterplot them first in their raw coordinates, then with one axis logarithmic and finally with both axes logarithmic. If it looks straight in the first plot, you try a linear model; in one of the semi-logarithmic plots, an exponential model; in the bi-logarithmic plot, a power model.
On this chart, the US curve is frighteningly straight, with the number of deaths doubling in slightly less than 3 days. China’s curve is concave, meaning that it is no longer growing exponentially. The word “exponential” is loosely applied in the media to any kind of rapid growth but it has a precise, technical meaning that is honored on this chart: the next increment is proportional to the current, cumulative value, like compounding interest, and it’s why you can describe it in terms of the fixed time it takes for the value to double.
We pay attention to this chart because the information in it is of vital importance to us. Its design, however, is essential to our ability to understand it. It’s a long way from the pies and stacked bars that are the mainstay of factory performance boards.
#datavisualization
Contents
Share this:
Like this:
Related
By Michel Baudin • Press clippings • 0 • Tags: Data visualization