gmGeostats (version 0.10-6)

anaBackward: Backward gaussian anamorphosis backward transformation to multivariate gaussian scores

Description

Backward gaussian anamorphosis backward transformation to multivariate gaussian scores

Usage

anaBackward(
  x,
  Y,
  sigma0,
  sigma1 = 1 + sigma0,
  steps = 30,
  plt = FALSE,
  sphere = TRUE,
  weights = NULL
)

Arguments

x

matrix of gaussian scores to be back-transformed

Y

node points defining the transformation (a matrix, same nr of columns)

sigma0

starting spread of the kernels in the forward transform

sigma1

final spread of the kernels in the forward transform

steps

number of steps to linearize the transform (default 30 is good)

plt

boolean, do you want to get a plot of the transformation?

sphere

boolean, should the data be taken as pre-Y-spherified? defaults to true

weights

vector of weights for all computations, length must be equal to number of rows of x

Value

a matrix with the scores back-transformed to the same scale as Y; same dimensions of x

See Also

ana() for defining a function that carries over the transformation (by means of a closure), anaBackward() for the explicit back-transformation, sphTrans() for defining a function that carries over the spherification of the data

Examples

# NOT RUN {
data("jura", package="gstat")
Y = jura.pred[,c(10,12,13)]
plot(compositions::acomp(Y))
Ylr = compositions::alr(Y)
Xns = matrix(rnorm(500), ncol=2)
plot(Ylr)
points(Xns, col=2, pch=4)
Xlr = anaBackward(x=Xns, Y=Ylr, sigma0=0.1)
qqplot(Xlr[,1], Ylr[,1])
qqplot(Xlr[,2], Ylr[,2])
qqplot(Xlr[,1]+Xlr[,2], Ylr[,1]+Ylr[,2])
# }

Run the code above in your browser using DataCamp Workspace