mcmc: Runs MCMC or optimisation to estimate Rt, cases and reporting...
In ben18785/incidenceinflation: Estimate R0 Given Case Reporting Delays

View source: R/sampling.R

mcmc	R Documentation

Runs MCMC or optimisation to estimate Rt, cases and reporting parameters

Description

Runs MCMC or optimisation to estimate Rt, cases and reporting parameters

Usage

mcmc(
  niterations,
  data,
  priors,
  serial_parameters,
  initial_cases_true,
  initial_reporting_parameters,
  initial_Rt,
  reporting_metropolis_parameters = list(mean_step = 0.25, sd_step = 0.1),
  serial_max = 40,
  p_gamma_cutoff = 0.99,
  maximise = FALSE,
  print_to_screen = TRUE,
  nchains = 1,
  is_parallel = FALSE,
  initial_overdispersion = 5,
  is_negative_binomial = FALSE,
  overdispersion_metropolis_sd = 0.25
)

Arguments

`niterations`	number of MCMC iterations to run or number of iterative maximisations to run
`data`	a tibble with five columns: time_onset, time_reported, cases_reported, Rt_index, reporting_index
`priors`	a named list with: 'Rt', 'reporting', 'max_cases'. These take the form: 'Rt' is a named list with elements 'shape' and 'rate' describing the gamma prior for each Rt; 'reporting' is a named list with elements 'mean_mu', 'mean_sigma', 'sd_mu', 'sd_sigma' representing the gamma prior parameters for the mean and sd parameters of the reporting parameters (itself described by a gamma distribution); max cases controls the upper limit of the discrete uniform distribution representing the prior on true cases
`serial_parameters`	named list of 'mean' and 'sd' of gamma distribution characterising the serial interval distribution
`initial_cases_true`	a tibble with two columns: "time_onset" and "cases_true", which represents initial estimates of the true number of cases with each onset time.
`initial_reporting_parameters`	provides initial guesses of the mean and sd of the reporting delay distribution(s). These can be either a named with two named elements ('mean', 'sd') for a time-invariant reporting delay or a tibble with three columns: 'reporting_piece_index', 'mean', 'sd' (where the number of indices corresponds to the number provided in the data frame).
`initial_Rt`	initial guess of the Rt values in each of the piecewise segments. Provided in the form of a tibble with columns: 'Rt_index' and 'Rt'
`reporting_metropolis_parameters`	named list of 'mean_step', 'sd_step' containing step sizes for Metropolis step
`serial_max`	maximum point at which to truncate sum in renewal process
`p_gamma_cutoff`	a p value (0 <= p <= 1) indicating the threshold above which we deem certainty
`maximise`	rather than sample a case count give the case count with the maximum probability (by default is FALSE)
`print_to_screen`	prints progress of MCMC sampling to screen. Defaults to true. Disabled when is_parallel is TRUE.
`nchains`	number of Markov chains to run. Defaults to 1
`is_parallel`	Boolean to indicate whether or not to run chains in parallel. Defaults to FALSE.
`initial_overdispersion`	the initial value of the overdispersion parameter if assuming a negative binomial sampling model (default to 5).
`is_negative_binomial`	if negative-binomial renewal model specified (defaults to FALSE)
`overdispersion_metropolis_sd`	the standard deviation of the proposal kernel