search

maxbiostat / R0_uncertainty

Some remarks on prior modelling for the basic reproductive number in the Susceptible-Infected-Recovered (SIR) epidemic model
6 stars 2 forks source link

Compare choices for the likelihood #31

Open maxbiostat opened 4 years ago

maxbiostat commented 4 years ago

This issue deals with the choice of likelihood. Let the data be Y_t and and let the solution to the ode be mu(t).

We can either say that

Y_t ~ negative_binomial(mu(t), sigma)

or 

Y_t  ~ lognormal(log(mu(t)), sigma)

We should test (a) if these make a difference in terms of parameter estimates and ESSs and (b) whether one runs faster than the other.