Probabilistic program inference in network-based epidemiological simulations

Fig 4

Samples for the NSEIR model for Miami-Dade using parameters learned from likelihood weighting and Metropolis-Hastings.

Neither alternate method is able to produce a good fit.