stCopPredict: Spatio-Temporal Prediction based on a Spatio-Temporal Vine...

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

Description

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

Usage

stCopPredict(predNeigh, dataST, predST, stVine, margin, method = "quantile", p = 0.5, ...)

Arguments

predNeigh	the neighbourhood used for prediction, its slot prediction must be TRUE.
dataST	the same ST object as used in the generation of predNeigh providing the data for interpolation.
predST	the same ST object as used in the generation of predNeigh providing the prediction locations for interpolation.
stVine	the spatio-temporal vine copula describing the spatio-temporal 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 condStVine through numerical integration/optimisation.

Value

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

Author(s)

Benedikt Graeler

See Also

condStVine and spCopPredict for the spatial version.

Examples

library("sp")
library("spacetime")

spCopT0 <- spCopula(components=list(claytonCopula(8), claytonCopula(4), 
                                    claytonCopula(2), claytonCopula(1),
                                    claytonCopula(0.5), indepCopula()),
                    distances=c(0,100,200,300,400,500),
                    unit="km")
spCopT1 <- spCopula(components=list(claytonCopula(4), claytonCopula(2), 
                                    claytonCopula(1), claytonCopula(0.5),
                                    indepCopula()),
                    distances=c(0,100,200,300,400),
                    unit="km")

stCop <- stCopula(components=list(spCopT0, spCopT1),
                  tlags=-(0:1))

stVineCop <- stVineCopula(stCop, vineCopula(4L))

sp <- SpatialPoints(matrix(c(181000,181100,333500,333600),2))
time <- as.POSIXct("2014-03-18")+60*60*24*c(0,1,2)
data <- data.frame(var1=runif(6))

stData <- STFDF(sp, time, data)
stQuerry <- STF(SpatialPoints(matrix(c(181000,181200,333600,333600),2)),
                time[2:3])

stNeigh <- getStNeighbours(stData=stData, ST=stQuerry, prediction=TRUE, spSize=3,
                           tlags=-(0:1))

stCopPredict(stNeigh, stData, stQuerry, stVineCop, list(q=qunif), "quantile", 0.5)

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