traceKrigeCV: Performs k-fold cross validation on the kriging model:

Usage Arguments Details Author(s)

Usage

1
traceKrigeCV(.coordinates, .functions, .model, .F = NULL)

Arguments

Coordinates

- Coordinates of your functional data

Functions

- Functions stacked in columns

Variogram

Model - Class vgm, produced as a call to gstat::vgm

.covType

- Covariance type: "omni", "product", "sum".

F

- Universal kriging matrix. For ordinary kriging a vector of 1's. For simple kriging NULL which is default!

Details

This function performs k-fold cross validation for a given variogram model If F=NULL it assumes simple kriging in which case it must work with covariances. Covariances are produced from the variogram by substracting the variogram values from the fitted sill. Of course, this assumes second order stationarity! It's users responsibility to verify that this is truly appropriate!

In all other cases of universal kriging (OK as a special case), the code will work with variograms.

Note: This is an experimental function! Use at your own responsibility!

Author(s)

Ogy Grujic (ogyg@stanford.edu)


ogru/fdagstat documentation built on May 29, 2019, 7:19 a.m.