covid-19

These are the results of my analysis. I have used an Exponential and a Logistic model. I plot the best fit model and also its gradient. I minimize with Minuit, and then I run a MCMC (emcee). The vertical blue lines show the fit range interval. MCMC quantiles are (0.16, 0.5, 0.84). Here I investigate only deaths for Italy.

First, I fit data up to March the 9th, and I project the model, both for Logistic and Exponential. This allows to have a better feeling of the prediction power of the two models. Logistic seems to adapt better, but due to several issues, in principal the migration of people to south last week, we should wait a few days to draw firm conclusions. ( see figures titles for details) UPDATED to 15/03/2019

As second attempt, I fit both the Logistic and Exponential model over the full data, and predict an asymptotic valued of deaths of ~ 3264 at 1-sigma, with upper and lower error of ~ 300 and ~ 230, respectively . The trend seems to be more likely logistic (see residuals) with new data, but it is too early!

Here I investigate possible delays due to migration from North to South of Italy after North lockdown. The rest of Italy (orange and green lines) population trend is rescaled to North (blue lines) trend, rescaling by respective population ratio. The green line is the orange one, moved backward a week. Apparently, rest of Italy is following the trend in the North, with a delay of one week.