COVID-19 in NYC - Analysis of Latest Data

From my latest NYC analysis, (cell 12, in https://github.com/hjstein/coronavirus-data/blob/master/Notebooks/Current.ipynb, updated this morning), it looks like the NYC growth rate slowed substantially on December 5th, about 1.5 weeks after Thanksgiving:

The growth rate is close to zero, but is still positive.  And we now have almost 225 hospitalizations/day and close to 30 deaths/day.

As of this morning, 12/21, the reports only go through 12/17, so the numbers only look accurate up to around 12/11, at which point we were up to ~2,600 infections/day (on a 7 day rolling average basis).  I expect the more recent numbers could get to over 2,800:

Cell 14 shows that we're experiencing over 220 hospitalizations/day now, and will probably breach 225.

Extrapolating based on cell 16, it looks like we might have surpassed 30 deaths/day:

(For an explanation of the above graphs, see previous blog posts.)

It makes sense that after the big events of the fall (Halloween, Thanksgiving, etc), that we'd see a large increase in infections.  After these events, I had expected infection rates to come down about a 1-2 weeks later.  What we've seen is that after each of these events, the levels have stabilized but not dropped.

In retrospect, this make sense.  Consider R, which is the number of people that an infected person infects.  When R>1, the number of cases will rise over time, and when R<1, the number of cases will drop.  Through July & August, the cases per day has been flat, indicating an R of about 1.  Each of these events leaves a larger percentage of the population infectious, so if the mitigation actions (lock downs, distancing, masks, ...) aren't changed, then you have a larger infectious population with the same R=1 value, so the numbers can remain high and won't drop until additional actions are taken, like identifying and isolating cases, or stopping the gatherings that lead to the most infections.