predonfly: Probability forecasting

In subrank: Computes Copula using Ranks and Subsampling

Probability forecasting

Description

From two sets of observations, first one of complete observations and second one of incomplete observations, provides simulated values of the unknown coordinates.

Usage

predonfly(completeobs,incompleteobs,varnames,subsampsize,nbpreds=1,mixties=FALSE,
          maxtirs=1e5,complete=TRUE,nthreads=2)

Arguments

completeobs	the set of complete observations.
incompleteobs	the set of incomplete observations.
varnames	the modeled variables.
subsampsize	the sub-sample size.
nbpreds	the number of predictions for each incomplete observation.
mixties	if TRUE, should put equal weight on tied values, using random permutations (not yet implemented)
maxtirs	the maximum number of sub-samples, to stop the computation even if they did not provide nbpreds predictions for each incomplete observation.
complete	If TRUE, predictions are completed with incomplete observations
nthreads	number of number of threads, assumed to be strictly positive. For "full throttle" computations, consider using parallel::detectCores()

Value

the matrix of the completed observations

Author(s)

Jerome Collet

Examples

lon=100
plon=30
subsampsize=10

x=rnorm(lon)
y=2*x+rnorm(lon)*0
donori=as.data.frame(cbind(x,y))
##
knownvalues=data.frame(x=rnorm(plon))
prev <- predonfly(donori,knownvalues,c("x","y"),subsampsize,100)

##
plot(prev$x,prev$y,pch=20,cex=0.5,
     ylim=range(c(prev$y,donori$y),na.rm=TRUE),xlim=range(c(prev$x,donori$x)))
points(donori[,1:2],col='red',pch=20,cex=.5)

lon=3000
mg=20
dimtot=4
rayon=6

genboules <- function(lon,a,d)
{
  ss <- function(vec)
  {return(sum(vec*vec))}
  surface=matrix(nrow=lon,ncol=d,data=rnorm(lon*d))
  rayons=sqrt(apply(surface,1,ss))
  surface=surface/rayons
  return(matrix(nrow=lon,ncol=d,data=rnorm(lon*d))+a*surface)
}

##############

donori=genboules(lon,rayon,dimtot)
donori=as.data.frame(donori)

dimconnues=3:dimtot
valconnues=matrix(nrow=1,ncol=length(dimconnues),data=0)
valconnues=as.data.frame(valconnues)
names(valconnues)=names(donori)[3:dimtot]
prev <- predonfly(donori,valconnues,names(donori),subsampsize,100)

boule2=genboules(plon,rayon,2)

plot(boule2[,1:2],xlab='X1',ylab='X2',pch=20,cex=.5)
plot(prev$V1,prev$V2,xlab='X1',ylab='X2',pch=20,cex=.5)

subrank documentation built on April 6, 2023, 1:11 a.m.