pre.sparsereg3D: Prepearing data for model selection ('sparsereg3D.sel'...

Description Usage Arguments Value

View source: R/pre.sparsereg3D.r

Description

Prepearing data for model selection (sparsereg3D.sel function) and model evaluation (sparsereg3D.ncv function)

Usage

1
2
3
4
pre.sparsereg3D(response, base.model, profiles, cov.grids, use.hier = FALSE,
  poly.deg = 1, num.folds = 10, num.means = 3, use.interactions = TRUE,
  standardize = TRUE, coord.trend = TRUE, seed = 321, kmean.vars = NULL,
  cum.prop = 0.9, s = 1)

Arguments

response

Only for compositional data modeling. Default is NA. "response" specifies the names of variables in compositions. For example "response = c("sand","silt","clay")"

base.model

model description (class "formula" of form "target.variable ~ covariates + depth" or in case of compositional data modeling just "~ covariates + depth", without target variable specified in the formula.)

profiles

observations of target variable (class "SoilProfileCollection")

cov.grids

covariates (class SpatialPixelsDataFrame)

use.hier

logical. If TRUE hierarchy constraints will be enforced.

poly.deg

degree of polynomial depth function

num.folds

number of folds in crossvalidation

num.means

number of clusters in k-means clustering

use.interactions

logical. If TRUE interactions will be included in the model.

standardize

logical. If TRUE standardization will be performed.

s

Integer. Only for compositional data modeling. "s" specifies which variable in composition will be used as target variable for stratification. For example, s = 1 in composition "sand, silt, clay" specifies "sand".

Value

List of objects including:

@keywords preprocessing


pejovic/sparsereg3D documentation built on May 25, 2019, 12:45 a.m.