Specify modskurt model for the distribution of `y` over `x`

Usage

mskt_spec(
  data,
  y,
  x,
  effort = NULL,
  dist = "nb",
  shape = "rdp",
  hyperparams = list(),
  pred_grid = NULL,
  subset_prop = 1,
  subset_seed = NULL
)

Arguments

data: data.frame or tibble
y: species count, optionally named - see details
x: environmental gradient, optionally named - see details
effort: sampling effort used to observe y
dist: distribution of y over x
shape: shape of distribution mean - see ?mskt_shape()
hyperparams: list of hyper-parameters to adjust in-built priors
pred_grid: vector of x values to make y predictions over
subset_prop: proportion of data to use for early model checks
subset_seed: a seed for the subset sample proportion

Value

list of standata for prior verification and posterior estimation

Details

## Forms of the discrete (count) data distribution - `"nb"`: Negative Binomial - `"zinbl"`: Zero-inflated Negative Binomial (Linked)

An interactive graph of these count distributions and possible parameters is available at [https://salt-ecology.shinyapps.io/nb-zinbl-prior-specs/]().

## Shapes for the mean function - default: height (H), mode (m), scale (s) - optional extras + `"r"` assymmetry or skew + `"d"` flatness or peakedness + `"p"` tail exaggeration or pinching

These can be combined as desired, e.g. `"rp"`, `"rdp"`, ...

An interactive graph of the modskurt mean function and possible parameters is available at [https://salt-ecology.shinyapps.io/modskurt-prior-specs/]().