inlabru: lgcp – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

lgcp

Log Gaussian Cox process (LGCP) inference using INLA

Description

This function performs inference on a LGCP observed via points residing possibly multiple dimensions. These dimensions are defined via the left hand side of the formula provided via the model parameter. The left hand side determines the intensity function that is assumed to drive the LGCP. This may include effects that lead to a thinning (filtering) of the point process. By default, the log intensity is assumed to be a linear combination of the effects defined by the formula's RHS. More sofisticated models, e.g. non-linear thinning, can be achieved by using the predictor argument. The latter requires multiple runs of INLA for improving the required approximation of the predictor. In many applications the LGCP is only observed through subsets of the dimensions the process is living in. For example, spatial point realizations may only be known in sub-areas of the modelled space. These observed subsets of the LGCP domain are called samplers and can be provided via the respective parameter. If samplers is NULL it is assumed that all of the LGCP's dimensions have been observed completely.

Usage

lgcp(
  components,
  data,
  samplers = NULL,
  domain = NULL,
  ips = NULL,
  formula = . ~ .,
  E = NULL,
  options = list()
)

Arguments

`components`	A formula describing the latent components
`data`	A data frame or `SpatialPoints(DataFrame)` object
`samplers`	A data frame or `Spatial[Points/Lines/Polygons]DataFrame` objects
`domain`	Named list of domain definitions
`ips`	Integration points (overrides `samplers`)
`formula`	If NULL, the linear combination implied by the `components` is used as a predictor for the point location intensity. If a (possibly non-linear) expression is provided the respective Taylor approximation is used as a predictor. Multiple runs if INLA are then required for a better approximation of the posterior.
`E`	Single numeric used rescale all integration weights by a fixed factor
`options`	See `bru_options_set()`

Value

An bru() object

Examples

if (bru_safe_inla()) {

  # Load the Gorilla data
  data(gorillas, package = "inlabru")

  # Plot the Gorilla nests, the mesh and the survey boundary
  ggplot() +
    gg(gorillas$mesh) +
    gg(gorillas$nests) +
    gg(gorillas$boundary) +
    coord_fixed()

  # Define SPDE prior
  matern <- INLA::inla.spde2.pcmatern(gorillas$mesh,
    prior.sigma = c(0.1, 0.01),
    prior.range = c(0.01, 0.01)
  )

  # Define domain of the LGCP as well as the model components (spatial SPDE
  # effect and Intercept)
  cmp <- coordinates ~ mySmooth(map = coordinates, model = matern) + Intercept

  # Fit the model (with int.strategy="eb" to make the example take less time)
  fit <- lgcp(cmp, gorillas$nests,
    samplers = gorillas$boundary,
    domain = list(coordinates = gorillas$mesh),
    options = list(control.inla = list(int.strategy = "eb"))
  )

  # Predict the spatial intensity surface
  lambda <- predict(fit, pixels(gorillas$mesh), ~ exp(mySmooth + Intercept))

  # Plot the intensity
  ggplot() +
    gg(lambda) +
    gg(gorillas$mesh) +
    gg(gorillas$nests) +
    gg(gorillas$boundary) +
    coord_fixed()
}

inlabru

Bayesian Latent Gaussian Modelling using INLA and Extensions

v2.3.1

GPL (>= 2)

Authors

Finn Lindgren [aut, cre, cph] (<https://orcid.org/0000-0002-5833-2011>, Finn Lindgren continued development of the main code), Fabian E. Bachl [aut, cph] (Fabian Bachl wrote the main code), David L. Borchers [ctb, dtc, cph] (David Borchers wrote code for Gorilla data import and sampling, multiplot tool), Daniel Simpson [ctb, cph] (Daniel Simpson wrote the basic LGCP sampling method), Lindesay Scott-Howard [ctb, dtc, cph] (Lindesay Scott-Howard provided MRSea data import code), Seaton Andy [ctb] (Andy Seaton provided testing and bugfixes)

Initial release