predict.Aniso: Obtain predictions at unobserved locations for the stationary...

In markdrisser/convoSPAT: Convolution-Based Nonstationary Spatial Modeling

Description

predict.Aniso calculates the kriging predictor and corresponding standard errors at unmonitored sites.

Usage

## S3 method for class 'Aniso'
predict(object, pred.coords, pred.covariates = NULL,
  pred.fixed.nugg2.var = NULL, ...)

Arguments

object	An "Aniso" object, from Aniso_fit.
pred.coords	Matrix of locations where predictions are required.
pred.covariates	Matrix of covariates for the prediction locations, NOT including an intercept. The number of columns for this matrix must match the design matrix from mean.model in NSconvo_fit. Defaults to an intercept only.
pred.fixed.nugg2.var	An optional vector or matrix describing the the variance/covariance a fixed second nugget term (corresponds to fixed.nugg2.var in Aniso_fit; often useful if conducting prediction for held-out data). Defaults to zero.
...	additional arguments affecting the predictions produced.

Value

A list with the following components:

pred.means	Vector of the kriging predictor, for each location in pred.coords.
pred.SDs	Vector of the kriging standard errors, for each location in pred.coords.

Examples

## Not run: 
pred.S <- predict( Aniso.obj,
pred.coords = cbind(runif(300),runif(300)) )

## End(Not run)

markdrisser/convoSPAT documentation built on May 21, 2019, 11:48 a.m.