Optimize sample allocation across strata

Calls the sightability API to determine optimal allocation of survey samples using Neyman allocation.

Usage

sight_optimize_design(
  method,
  value,
  species,
  survey_type,
  analysis_unit,
  bio_year,
  samp_var = NULL,
  sight_var = NULL,
  mod_var = NULL,
  confidence = 0.9,
  design_overrides = NULL,
  filter_species = TRUE,
  filter_survey_type = TRUE,
  ...
)

sight_optimize_design_id(
  method,
  value,
  species_id,
  survey_type_id,
  analysis_unit_id,
  bio_year,
  samp_var = NULL,
  sight_var = NULL,
  mod_var = NULL,
  confidence = 0.9,
  design_overrides = NULL,
  dry_run = FALSE,
  ...
)

Arguments

method: Character. Optimization method: "precision" (target error bound) or "fixed" (target sample size).
value: Numeric. For "precision": desired error bound. For "fixed": target sample size.
species: Character. Species name, required for looking up survey_type or analysis_unit by name.
survey_type: Character. Survey type name.
analysis_unit: Character. Analysis unit name.
bio_year: Integer. Biological year for the survey design.
samp_var: Numeric. Optional sampling variance from a previous model fit.
sight_var: Numeric. Optional sightability variance from a previous model fit.
mod_var: Numeric. Optional model variance from a previous model fit.
confidence: Numeric. Confidence level for the error bound (default 0.90 for 90% CI). Must be between 0 and 1.
design_overrides: A list of per-stratum overrides. Each element is a list with stratum_id (required) and optional subunits_available, pop_sd, and/or cost values. NULL (default) means no overrides.
filter_species: logical, if FALSE species is only used for lookup context and not passed as a filter (default TRUE)
filter_survey_type: logical, if FALSE survey_type is only used for lookup context and not passed as a filter (default TRUE)
...: Additional arguments passed to the underlying spdgt.auth function (e.g., verbosity, timeout).
species_id: Integer. Species identifier.
survey_type_id: Integer. Survey type identifier.
analysis_unit_id: Integer. Analysis unit identifier.
dry_run: If TRUE, print the HTTP request without sending it and return the request object invisibly. Useful for debugging.

Value

Tibble with one row per stratum and the following columns:

method: Character. The optimization method used ("precision" or "fixed").
value: Integer. The input value (error bound for "precision", sample size for "fixed").
species_id: Integer. Species identifier.
survey_type_id: Integer. Survey type identifier.
analysis_unit_id: Integer. Analysis unit identifier.
bio_year: Integer. Biological year.
stratum_id: Integer. Stratum identifier (excludes stratum 0 / "Other").
subunits_available: Integer. Total subunits available in the stratum.
pop_sd: Numeric. Population standard deviation for the stratum, estimated from historical survey data.
cost: Integer. Cost per sample unit (typically 1).
sample_proportion: Numeric. Optimal Neyman allocation proportion for this stratum (sums to 1 across included strata; NA for excluded strata).
n_samples: Integer. Number of subunits to sample in this stratum.
dau_proportion: Numeric. Sampling intensity (n_samples / subunits_available).
include: Logical. Whether the stratum was included in the optimization.

Examples

if (FALSE) { # \dontrun{
# Using names
sight_optimize_design(
  method = "fixed",
  value = 100,
  species = "Mule Deer",
  survey_type = "Sightability-Abundance",
  analysis_unit = "Bannock",
  bio_year = 2024
)

# Using IDs with variance from previous model
sight_optimize_design_id(
  method = "precision",
  value = 500,
  species_id = 1,
  survey_type_id = 3,
  analysis_unit_id = 272,
  bio_year = 2024,
  samp_var = 15000,
  sight_var = 2000,
  mod_var = 800
)
} # }

Usage

Arguments

Value

See also

Examples