Create control parameters

Create control parameters

chronofix_mcmc_control(
  n_steps = 1000,
  burnin = 0,
  thinning_factor = 1,
  n_chains = 1,
  parallel = FALSE,
  n_workers = 1,
  lower_quantile = 0.01,
  upper_quantile = 0.99,
  earliest_possible_date = NULL,
  latest_possible_date = NULL,
  date_buffer = 30,
  mean_sdlog = 1,
  cv_sdlog = 1,
  prob_update_estimated_dates = 0.1,
  prob_update_error_indicators = 0.1,
  prob_error_swap = 1,
  cascade_sampling = FALSE
)

Arguments

n_steps

The number of steps to run in each MCMC chain

burnin

The number of steps at the beginning of each chain to discard as burnin

thinning_factor

A thinning factor applied to the chains. If given, everythinning_factor'th step is retained

n_chains

The number of chains to run

parallel

Logical, indicating whether or not to run chains in parallel

n_workers

Number of workers to use for parallelisation

lower_quantile

Lower quantile used for initialisation of true dates

upper_quantile

Upper quantile used for initialisation of true dates

earliest_possible_date

The earliest possible date (in YYYY-MM-DD format). Dates will not be estimated as occurring earlier than this date. If NULL (the default), then it will be determined by the earliest date in the data and date_buffer.

latest_possible_date

The latest possible date (in YYYY-MM-DD format). Dates will not be estimated as occurring later than this date. If NULL (the default), then it will be determined by the latest date in the data and date_buffer.

date_buffer

The date buffer in terms of days to determine the earliest and/or latest possible dates from the data. If earliest_possible_date is not specified then the earliest possible date will be taken as date_buffer days before the earliest date in the data. Similarly, if latest_possible_date is not specified then the latest possible date will be taken as date_buffer days after the latest date in the data.

mean_sdlog

The sdlog proposal parameter for the delay means

cv_sdlog

The sdlog proposal parameter for the delay coefficients of variation

prob_update_estimated_dates

The probability of proposing an update to each estimated date at each iteration in the MCMC

prob_update_error_indicators

The probability of proposing an update to each error indicator (with the corresponding estimated date updated accordingly) at each iteration in the MCMC

prob_error_swap

The probability of proposing to swap all errors to non-errors and vice versa (excluding missing dates) for individuals with at least one error and non-error at each iteration in the MCMC

cascade_sampling

Logical, indicating whether or not to use cascade sampling for estimated dates when updating an error indicator

Value

List of control parameters