An in-development R package and a Bayesian hierarchical model jointly fitting multiple "local" wastewater data streams and "global" case count data to produce nowcasts and forecasts of both observations
Currently we have a helper function that adds a bit of noise to the global R(t) estimate if a user specifies a smooth set of weekly R(t) values.
Its possible this could create negative R(t) values plus we shouldn't really be adding noise in natural scale, so really the noise should be added in log scale, even if specified in natural scale and then approximated.
Goal
Currently we have a helper function that adds a bit of noise to the global R(t) estimate if a user specifies a smooth set of weekly R(t) values.
Its possible this could create negative R(t) values plus we shouldn't really be adding noise in natural scale, so really the noise should be added in log scale, even if specified in natural scale and then approximated.
https://github.com/CDCgov/ww-inference-model/pull/29/files#r1727133012