Make Data Relevant

All of the data I’ve seen displayed on this pandemic is lacks in a couple of key areas. First this pandemic is evolving over time and the graphs and plots need to make those trends clear and project forward so that people can see what their life might be like in the future. It is almost criminal that the CDC, CA department of health, and the SF gov websites treat the case numbers as completely static. If you look at these websites, all you see is a fixed number. You can’t see the trend over time and you have no way of understand what will happen at what point of time in the future.

The other way in which the facts are presented horribly is that they need to be presented much more locally. COVID-19 is very much a community infection and the data about the trends needs to be assessed on the community level. As you can see below there are huge differences in the trend lines between San Francisco and the nation. I attribute this in part to our Mayor, London Breed’s, leadership

The above plot is done on a logarithmic y scale axis so that exponential growth shows up as a straight line. This presentation style helps get around the human linear versus exponential bias, but notice the values on the y axis start as 10⁰ which is 1 and go up. 10¹ which is 10, 10² which is 100, and so on representing 1000, 10,000, 1000,000, and 1,000,000. This style of plot has another advantage in that the local community, the state and the national region all fit on one plot.

It is remarkable how well these lines fit exponential growth. In my data scientist experience I rarely see data that fits a model so nicely. The lines fitted above have the following doubling times.

Location	Doubling Period
San Francisco	3.79 days
California	2.99 days
United States	1.76 days

I have no idea why the US growth rate is so crazy high but it is an insane growth rate.

The doubling time is time it takes for a population to double in size/value. It is applied to population growth, inflation, resource extraction, consumption of goods, compound interest, the volume of malignant tumours, and many other things that tend to grow over time.

— wikipedia