COVID Projections:

When looking at COVID fatality data, 2 things jumped out at me. 1) correlation between state population density and deaths per capita and 2) the shape of the curve of fatalities. If you build a model that estimates per capita fatalities using pop density and weeks beyond peak fatality day, you get a very strong predictor (89% R^2).

The (linear) equation is:
Est fatalities per million = -56 + 14 * (weeks from peak) + .92 * (Pop Density)

Where there are large differences between estimated and actual, one could hypothesize that they are an indication of where that state is in the “curve” or lifecycle of the outbreak.

States that are the farthest along in the curve (expect declining fatality counts) are as follows:

States that are the furthest behind (expect rising fatality counts) are as follows: