That means we have to hope there are really millions infected. If the # is correct, this thing is batting .500 plus. I'm assuming recovered includes non symptomatics who test positive then after a while test negative.
Nobody reports "resolved cases". If you are sick with it, then recover, nobody is calling authorities to proclaim Joe Blow has now recovered. Case numbers is the denominator. Ignore the "resolved cases".
A "resolved (or 'recovered') case" requires 2 negative tests in a row over a 48 hour period.
I guarantee that's only being done for patients who are hospitalized with severe symptoms and wind up getting follow up tests as part of a discharge protocol after they've improved. . they aren't wasting those tests on your average Joe Blow with mild symptoms, thus Joe Blow will never get counted as "resolved."
Let us start by denoting with S(t), I(t), R(t), D(t), the number of susceptible, infected, recovered and dead persons respectively at time t in the population of size N. For our analysis, we assume that the total number of the population remains constant. Based on the demographic data for the province of Hubei N = 59m. Thus, the discrete SIRD model reads:
The above system is defined in discrete time points t = 1, 2, …, with the corresponding initial condition at the very start of the epidemic: S(0) = N − 1, I(0) = 1, R(0) = D(0) = 0. Here, β and γ denote the “effective/apparent” per day recovery and fatality rates. Note that these parameters do not correspond to the actual per day recovery and mortality rates as the new cases of recovered and deaths come from infected cases several days back in time. However, one can attempt to provide some coarse estimations of the “effective/apparent” values of these epidemiological parameters based on the reported confirmed cases using an assumption and approach described in the next section.
However, one can attempt to provide some coarse estimations of the “effective/apparent” values of these epidemiological parameters based on the reported confirmed cases using an assumption and approach described in the next section.
I'll be glad to do the grading, however I think I need to clarify/modify the formula given by EJD.
Restatement 1, add paren after deaths: Grading formula is (difference between prediction and actual number of cases) + (difference between prediction and actual number of deaths) x 25
First, his example, the score is 1000000 -1150000 + (10000 -12000) x 25 = -200,000.
Restatement 2, use absolute value of the differences so score is always non-negative: Grading formula thus is abs(difference between prediction and actual number of cases) + abs(difference between prediction and actual number of deaths) x 25
Deadline is Thursday 11:59PM CT, thus 9:59PM PT tonight, 12:59AM ET, Friday morning.