Curve estimation of the relationship between incidence and time in NYC show a linear relationship accounts for for 46% of variance -- in other words, looking at new case count as if it were a daily increase is not a great model. A cubic relationship accounts for about 81% of variance -- much better! Even more so if you don't accept NYS data from yesterday -- NYS reports a big jump in positive tests for NYC from the previous day, where NYC does not. Not sure why the discrepancy.
Anyway, here it is with the NYS data from yesterday (a rise in new cases at the right tail). Days elapsed are since March 2.
Everyone has a hobby
