project_from_filter: Project from Filtering Distribution

In zjiang2/haitipkg: R Package for Haiti Article

Project from Filtering Distribution

Description

This function simulates from a Haiti-cholera POMP model into the future starting from filtering distribution at the the last time point available.

Usage

project_from_filter(mod, end_states, covarGen = NULL, nsims = 100, seed = 123)

Arguments

mod	POMP model that you would like to simulate from.
end_states	The starting states from which to simulate. This is designed so that these states are draws from the filtering distribution at time $N$, but they can be any desired state from which to start the simulation.
covarGen	function that is used to generate covariates, if the model needs covariates to run rprocess.
nsims	Integer number of simulations desired.
seed	seed for the random number generator.

Details

Because we are primarily interested in projecting the number of Cholera cases in the future under various vaccination scenarios, it makes sense to use as much information as possible, as simulating from the start of the time series will ultimately disregard the known truth at the end of the available date. The state of the system at the end of the time series can be accounted for by simulating the filtering distribution:

Xk | Y1 = y1*, \ldots, Yk*

As currently implemented, the efficiency could be greatly improved for projecting models that do not have covariates.

zjiang2/haitipkg documentation built on March 1, 2024, 11:34 p.m.