extractSTA: Extract statistics from fitted models

View source: R/extractSTA.R

extractSTAR Documentation

Extract statistics from fitted models

Description

Extract and calculate various results for fitted models such as BLUEs, BLUPs, unit errors and heritabilities. For a full list of results that can be extracted, see the table below.

The result(s) to extract can be specified in what.
If a single result is extracted, this result is returned as a data.frame, If this is not possible, because the format of the result is incompatible with the data.frame format, the result is returned as a list. E.g. if BLUEs are extracted, the output of the function is a data.frame with BLUEs. However if varCompF is extracted, the output of the function is a list.

Results that are returned as data.frame are marked as such in the asDataFrame column in the table. If the default return value for a result is a data.frame that can be overridden by the user by specifying asDataFrame = FALSE. When doing so the result will be returned as a list of data.frames, one per trial. They other way round is not possible. If a result is returned as a list according to the table, it cannot be returned as a data.frame.
If multiple results are extracted at the same time, these are always returned as a list.

Most results can only be calculated if a model is fitted with genotype as fixed or with genotype as random. E.g. to compute heritabilities a model should be fitted with genotype as random effect. This is indicated in the table in the column model with "F" and "R" respectively.

Possible options for what are:

result model description asDataFrame
BLUEs F Best Linear Unbiased Estimators yes
seBLUEs F standard errors of the BLUEs yes
ue F unit errors - only for lme4 and asreml yes
varCompF F variance components for the model with genotype as fixed component
fitted F fitted values for the model with genotype as fixed component yes
residF F residuals for the model with genotype as fixed component yes
stdResF F standardized residuals for the model with genotype as fixed component yes
wald F results of the wald test - only for lme4 and asreml
CV F Coefficient of Variation yes
rDfF F residual degrees of freedom for the model with genotype as fixed component yes
sed F standard error of difference - only for asreml
lsd F least significant difference - only for asreml
BLUPs R Best Linear Unbiased Predictors yes
seBLUPs R standard errors of the BLUPs yes
heritability R generalized heritability yes
varCompR R variance components for the model with genotype as random component
varGen R genetic variance component yes
varErr R residual variance component yes
varSpat R spatial variance components - only for SpATS
rMeans R fitted values for the model with genotype as random component yes
ranEf R random genetic effects yes
residR R residuals for the model with genotype as random component yes
stdResR R standardized residuals for the model with genotype as random component yes
rDfR R residual degrees of freedom for the model with genotype as random component yes
effDim R effective dimensions - only for SpATS
ratEffDim R ratios of the effective dimensions - only for SpATS

Usage

extractSTA(
  STA,
  trials = names(STA),
  traits = NULL,
  what = "all",
  asDataFrame = length(what) == 1 && what != "all",
  keep = NULL,
  restoreColNames = FALSE
)

Arguments

STA

An object of class STA.

trials

A character vector of trials for which the statistics should be computed. If not supplied, statistics are computed for all trials that have been modeled.

traits

A character vector of traits for which the statistics should be computed. If not supplied, statistics are computed for all traits that have been modeled.

what

A character vector indicating which statistics should be computed. Most statistics are available for all models, some only for models fitted using a certain engine. If this is the case, this is indicated in the list with options in details.
If what = "all", all available statistics are computed.

asDataFrame

Should the output be reshaped to a data.frame. This is only possible if the number of statistics to extract is one.

keep

A character vector of column(s) in the object of class TD used for modeling. These columns will be kept as output when computing fitted values, residuals, standardized residuals and rMeans. Columns can also be kept when computing (se)BLUEs and (se)BLUPs but only if the column to keep contains unique values for the modeled variables, i.e. a column repId with several different values per genotype cannot be kept.

restoreColNames

Should the original column names be restored in the output of the extracted data?

Value

Depending on the input either a data.frame or a list with, per trial for which statistics have been extracted, a list of those statistics.

See Also

fitTD

Examples

## Fit model using SpATS.
modSp <- fitTD(TD = TDHeat05,
               design = "res.rowcol",
               traits = "yield")

## Extract all available statistics from the fitted model.
extr <- extractSTA(modSp)

## Extract only the BLUEs from the fitted model.
BLUEs <- extractSTA(modSp,
                    what = "BLUEs")

## Extract only the BLUEs from the fitted model and keep trial as variable in
## the output.
BLUEs2 <- extractSTA(modSp,
                     what = "BLUEs",
                     keep = "trial")


statgenSTA documentation built on Oct. 14, 2024, 9:10 a.m.