Written by Rahul Dave on 

Making the @FT Covid Vaccine Visualization

  1. Two weeks back, in the thread https://twitter.com/rahuldave/status/1425200284301725708, I talked about the visualization in this image from the financial times (@FT) https://www.ft.com/content/0f11b219-0f1b-420e-8188-6651d1e749ff?hcb=1 . Here I tackle the production of such an image in @matplotlib .

  1. And in another thread on Keeping it Simple: https://twitter.com/rahuldave/status/1427705386434392067, I talked about some of the ins-and-outs of @matplotlib, in the context of removing chartjunk from images. Here we'll use those techniques and some new ones...

  2. I used Engauge Digitizer, to extract the data points from the @FT graph. And the built-in OSX utility, Digital Color Meter, to figure the colors used there. The data and the notebook used to produce this thread are available at https://gist.github.com/rahuldave/df75d3aa73e95e8361b7cdd1663c9868.

  3. At first I made a simple image of the data, with the red points representing the unvaxed risk, and the blue the vaxed risk. I put a rough black line to compare the danger to an unvaxed 50 year old with that of a vaxed 80 year old.

  1. I then set up a grid like in the original plot, put labels everywhere, and get rid of external tick marks. You can read more about ticks at https://e2eml.school/matplotlib_ticks.html by @brohrer

  1. I formatted the ticks the way the @FT article did, and removed the spines that make up the plot bounding box as well.

  1. I replaced the horizontal line comparing ages with an arrow, as in the original image. The arrow is drawn as a Patch, which is an aribitrary 2-D region in matplotlib. Read more about Patches in @brohrer's course https://e2eml.school/matplotlib_patches.html and in the @matplotlib docs.

  1. At this point, our image is beginning to take shape. But we have data points for the vaxed case instead of arrows! I'll use Patches to fix this!

  1. By using patches I have now added an arrow for each age in our dataset! I also add a formatted title. We are getting closer to the final image. I need to annotate the plot and make it have the right color!

  1. I finally set both the figure and axes color and add 3 annotations. The annotations use arrows, which go from the xytext coordinates to the xy co-ordinates. The annotations serve as legends, and describe the main conclusion.

  1. I'll tackle more visualization stuff in future tweet threads, including the principles of good visualization and how to achieve them with @matplotlib . Follow me on twitter at @rahuldave and keep an eye on https://stories.univ.ai for more!

  2. Permalink to thread here: https://stories.univ.ai/makingthecovidviz/! If you'd like to learn Data Science and AI more formally, you might want to check out my company's programs at https://univ.ai/programs/, and the thread https://twitter.com/rahuldave/status/1424793241019428864 (discounts inside).