censored_index_fun: Compute Poisson-lognormal model-based indices of relative...

In pbs-assess/hookCompetition: An R package for estimating relative abundance indices from longline data using a censored likelihood approach

Compute Poisson-lognormal model-based indices of relative abundance See the vignette for details

Description

Compute Poisson-lognormal model-based indices of relative abundance See the vignette for details

Usage

censored_index_fun(
  data,
  survey_boundaries,
  species,
  M = 1000,
  return = T,
  ICR_adjust = F,
  cprop = 1.1,
  nthreads = 1,
  keep = F,
  use_upper_bound = FALSE,
  upper_bound_quantile = 1,
  plot = T,
  allyears = F,
  station_effects = T,
  prior_event = HT.prior(),
  prior_station = HT.prior(),
  init_vals = NULL,
  n_knots = 8,
  seed = 0,
  verbose = F,
  n_trajectories = 10,
  preserve_inter_regional_differences = F
)

Arguments

data	a sf points object containing the IPHC data
survey_boundaries	a sf polygons object containing the survey boundary definitions
species	a character name of the species (linked to the variables 'N_it' in the dataframe)
M	the number of independent Monte Carlo samples from the posterior used to compute indices
return	logical stating whether or not to return indices as a data.frame
ICR_adjust	A logical determining if the ICR-based scale-factor adjustment should be used
cprop	the minimum proportion of baits needed to be removed to induce censorship. If >1 no censorship.
nthreads	how many cores do you want to run in parallel when fitting the model?
keep	a logical stating if you want to return the INLA model object. Useful for inspecting DIC, properties etc.,
use_upper_bound	a logical stating if right-censored or interval-censored response (with concervative upper bound derived using Baranov Catch equation) is desired.
upper_bound_quantile	a proportion stating which quantiles of observation to remove censorship from. If greater than 1, do not remove censorship from any observation
plot	logical. Do you want to return a ggplot of the indices along with the data.frame?
allyears	logical determining if upper bound quantile is computed uniquely for each year (if False) or over all years (if TRUE)
station_effects	logical stating if IID station-station random effects wanted
prior_event	a prior distribution for the event-event IID random effect standard deviation (see HT.prior for default half t_7(0,1) distribution)
prior_station	a prior distribution for the station-station IID random effect standard deviation (see HT.prior for default half t_7(0,1) distribution)
init_vals	initial values for fitting model. If NULL let INLA choose them.
n_knots	the number of knots used for the temporal 'spline'. Default is 8. More knots means a 'wigglier' temporal effect is allowed, at a risk of higher variance.
seed	the seed used for Monte Carlo sampling. Note that 0 means the result is non-reproducible, but the code can be significantly faster.
verbose	logical. If TRUE, print INLA diagnostics on the console as the model fits. Can help to diagnose convergence issues.
n_trajectories	integer specifying how many Monte Carlo sampled relative abundance indices to plot on a spaghetti plot. This can help with interpreting uncertainty. Suggested value 10.
preserve_inter_regional_differences	Logical if TRUE estimated inter-regional differences in mean abundance are shown at a cost of higher variance. Does not affect coastwide index.

pbs-assess/hookCompetition documentation built on April 27, 2023, 12:47 p.m.