transform_starting_values: Transform starting values to the link scale

View source: R/functions.R

transform_starting_valuesR Documentation

Transform starting values to the link scale

Description

Takes in starting guesses for parameter values, and maps them to the real valued link scale. Even parameters already on the real scale (such as means) are transformed, to ensure they are uniquely determined. Starting values for splines and covariates are always set to 0, as are starting values for parameters that aren't specified.

Usage

transform_starting_values(
  starting_values,
  a_choice,
  dist_choice,
  options,
  DF = list()
)

Arguments

starting_values

A named list including starting values for "mu", "sigma", "w" and "dist.par" (in the case of a non-poisson model)

a_choice

The choice of flight path distribution

dist_choice

The choice of observed count distribution

options

The list of model options that specify the number of broods, spline parameters, covariate formulas, etc.

DF

A data.frame which must be included when covariate formulas are specified, so that the correct number of parameters required by each formula can be determined. Calling extract_counts on this data.frame first to sanitize it is recommended

Value

A named list of starting parameter values, all on the link scale.


calliste-fagard-jenkin/GAI documentation built on July 26, 2022, 11:20 a.m.