spCopPredict: spatial prediction based on a spatial vine copula

In BenGraeler/spcopula: Copula Driven Analysis - Multivariate, Spatial, Spatio-Temporal

Description

A spatial vine copula is used to predict values at unobserved locations conditioned on observations of a local neighbourhood.

Usage

spCopPredict(predNeigh, dataLocs, predLocs, spVine, margin,
             method = "quantile", p = 0.5, ...)

Arguments

predNeigh	the neighbourhood used for prediction, its slot prediction must be TRUE.
dataLocs	some Spatial class providing the data for the prediction.
predLocs	some Spatial class providing the prediction locations.
spVine	the spatial vine copula describing the spatial dependence
margin	the marginal distribution as a list with entries named "d" for the density function (PDF), "q" for the quantile function and "p" for cumulative distribution function (CDF).
method	one of "quantile" or "expectation" denoting the type of predictor.
p	only used for the quantile predictor indicating the desired fraction the quantile should correspond to.
...	Further arguments passed to integrate.

Details

Predictions are done based on condSpVine through numerical integration/optimisation.

Value

A Spatial object of the same type as provided in the slot locations of the argument predNeigh.

Author(s)

Benedikt Graeler

References

Graeler, B. and E. Pebesma (2011): The pair-copula construction for spatial data: a new approach to model spatial dependency. Procedia Environmental Sciences (Vol. 7, pp. 206 - 211), Elsevier.

See Also

condSpVine

Examples

library("sp")

data("meuse.grid")
coordinates(meuse.grid) <- ~x+y
gridded(meuse.grid) <- TRUE

data("meuse")
coordinates(meuse) <- ~x+y
data("spCopDemo")

calcKTauPol <- fitCorFun(bins, degree=3)

spCop <- spCopula(components=list(normalCopula(), tCopula(),
                                  frankCopula(), normalCopula(), claytonCopula(),
                                  claytonCopula(), claytonCopula(), claytonCopula(),
                                  claytonCopula(), indepCopula()),
                  distances=bins$meanDists,
                  spDepFun=calcKTauPol, unit="m")

spVineCop <- spVineCopula(spCop, vineCopula(4L))

meuse$rtZinc <- rank(meuse$zinc)/(length(meuse)+1)

dataLocs <- meuse[1:4,]
predLocs <- meuse.grid[c(9:12,16:19,25:28),]
predMeuseNeigh <- getNeighbours(dataLocs, predLocs, 
                                5, "rtZinc", prediction=TRUE, min.dist=-1)

qMar <- function(x) {
  qlnorm(x,mean(log(meuse$zinc)),sd(log(meuse$zinc)))
}

predMedian <- spCopPredict(predMeuseNeigh, dataLocs, predLocs,
                           spVineCop, list(q=qMar), "quantile", p=0.5)

## Not run: 
  spplot(predMedian, "quantile.0.5", 
         sp.layout=list("sp.points", meuse, pch = 19, col = "red"),
         col.regions=bpy.colors())

## End(Not run)

BenGraeler/spcopula documentation built on Nov. 20, 2020, 4:07 p.m.