correctSpatialHeterogeneity: Correct spatial heterogeneity

In timflutre/rutilstimflutre: Timothee Flutre's personal R code

Correct spatial heterogeneity

Description

Use kriging to correct spatial heterogeneity in a plant field trial. Kriging (i.e. prediction) is performed per year. Then, the predicted responses that controls would have had if they had been planted everywhere are subtracted from the observed responses from the other genotypes.

Usage

correctSpatialHeterogeneity(
  dat,
  response,
  fix.eff = NULL,
  min.ctls.per.year = 10,
  cressie = TRUE,
  vgm.model = c("Exp", "Sph", "Gau", "Ste"),
  nb.folds = 5,
  out.prefix = NULL,
  verbose = 1
)

Arguments

dat	data frame with, at least, columns named "geno", "control" (TRUE/FALSE), "rank", "location", "year" and <response>
response	column name of dat corresponding to the response for which spatial heterogeneity will be corrected
fix.eff	if not NULL, vector of column names of data corresponding to fixed effects to control for in the kriging (e.g. "block")
min.ctls.per.year	minimum number of control data points in a given year to proceed
cressie	if TRUE, the variogram function from the gstat package uses Cressie's robust variogram estimate, else it uses the classical method of moments
vgm.model	type(s) of variogram model(s) given to the vgm function of the gstat package; if several, the best one (smaller sum of squared errors) will be used
nb.folds	number of folds for the cross-validation
out.prefix	prefix of the output files to save plots and results (if not NULL)
verbose	verbosity level (0/1/2)

Value

data frame as dat but with an additional column named <response>.csh

Author(s)

Timothee Flutre

timflutre/rutilstimflutre documentation built on Aug. 18, 2024, 7:43 p.m.