/sci/ - Science & Math

>>11522896
I'm trying to model the COVID-19 spread as a logistic curve.

At the moment I have a script that guesses how far we are along the logistic curve and draws it out based on that (for example, if we're halfway through the pandemic, then our peak will be at <current_n_cases> / 0.5). The script then does a binary search for the ratio that matches the best.

I have very little idea if what I'm doing is statistically sound. For example, Right now, my model says we peak at about 15M cases, but I'm not sure how much faith I should have. If I include some data from February, this number drops to like 4M (but again, should I be including data from two months ago?)
I know that you use R^2 as a statistical significance test in linear regression, but don't know how that applies to a logistic curve. Specifically, I'm not sure what my objective function should be.

Any and all help is appreciated.