buildMetadata {psgp}

Build Metadata Table

Description

buildMetadata builds a metadata table of likelihood model descriptors in the PSGP framework. This is an internal function and it should not be used directly.

Usage

  buildMetadata(observations)

Arguments

observations

an observations data frame containing a vector of observation variances (obserations$oevar) and, optionally, a list of observation biases (observationS$oebias). If the biases are omitted, a zero bias is assumed for all likelihood models.

Details

buildMetadata builds a metadata table of likelihood model descriptors in the PSGP framework. The likelihood models are assumed Gaussian with variances specified in the vector observations$oevar (the bias is assumed to be zero). Optionally, biases can be specified in the observations$oebias vector. However, biases are not taken into account in the current version of the PSGP package (they will be in a future release).

Author(s)

Remi Barillec

See Also

learnParameters, makePrediction, estimateParameters, spatialPredict,

Examples

  ## Load our favourite dataset
  data(meuse)
  obs <- meuse
  
  ## Number of observations
  nobs <- length(obs$y)
  
  ## Indicate which likelihood model should be used for each observation
  obs$oeid  <- seq(1:nobs)
  
  ## Use random variances for the sake of the example
  obs$oevar <- rnorm( max(obs$oeid) )
  
  ## Generate metadata table and print it out
  metadata <- buildMetadata(obs)
  print(metadata)