Skip to contents

Predict cover from a cover_fit

Source: R/cover_hurdle_diag.R
predict.cover_fit.Rd

Occurrence probability is always p = plogis(X * beta_occ). The conditional positive cover mu depends on the positive-part family:

Usage

# S3 method for class 'cover_fit'
predict(
  object,
  newdata = NULL,
  type = NULL,
  include_RE = FALSE,
  times = NULL,
  time_col = NULL,
  level = 0.95,
  nsim = 1000L,
  draws = TRUE,
  ...
)

Arguments

object

A cover_fit.

newdata

A data frame of covariates matching the original formula. For the nested-Laplace fit, one row per spatial unit (or a cell column).

type

Separate-Laplace fit: one of "expected", "occupancy", "conditional". Nested-Laplace fit: "occurrence", "cover_cond", "cover_exp", or "change" (the legacy aliases are accepted and mapped).

include_RE

Ignored for the separate-Laplace fit (no spatial projection); the nested-Laplace fit always projects the shared field.

times, time_col, level, nsim, draws

Nested-Laplace fit only: times = c(t1, t2) and time_col drive the "change" map; level is the credible level, nsim the draw count, draws whether to attach the draw matrices.

...

Unused.

Value

Separate-Laplace fit: a numeric vector. Nested-Laplace fit: a tobs_prediction.

Details

  • positive = "lognormal": mu = exp(eta_pos + sigma_pos^2 / 2) (lognormal back-transform on log-cover).

  • positive = "beta": mu = plogis(eta_pos) (mean of the beta on the natural cover scale with logit link).

Expected cover is E[y] = p * mu under both positive parts.

The separate-Laplace fit (method = "laplace") returns a fixed-effects-only numeric vector. The nested-Laplace shared-field fit (method = "nested_laplace") instead projects the shared occupancy-cover field and returns a tobs_prediction of posterior draws – the same tidy / change contract as predict.tobs_fit() for occu_cover(): pass type = "change" with times = c(t1, t2) and time_col for a per-cell delta map. Each prediction unit is a row of newdata (or a cell column indexing the field cells), and every quantity is marginalized per draw over the grid-integrated joint posterior.