var.select.md: Variable selection with Minimal Depth (MD)

In StephanSeifert/SurrogateMinimalDepth: Surrogate minimal depth variable importance

Variable selection with Minimal Depth (MD)

Description

This function executes MD applying ranger for random forests generation and is a reimplementation of var.select from randomForestSRC package.

Usage

var.select.md(
  x = NULL,
  y = NULL,
  ntree = 500,
  type = "regression",
  mtry = NULL,
  min.node.size = 1,
  num.threads = NULL,
  status = NULL,
  save.ranger = FALSE,
  create.forest = TRUE,
  forest = NULL,
  save.memory = FALSE,
  case.weights = NULL
)

Arguments

x	matrix or data.frame of predictor variables with variables in columns and samples in rows. (Note: missing values are not allowed)
y	vector with values of phenotype variable (Note: will be converted to factor if classification mode is used). For survival forests this is the time variable.
ntree	Number of trees. Default is 500.
type	Mode of prediction ("regression" or "classification"). Default is regression.
mtry	Number of variables to possibly split at in each node. Default is no. of variables^(3/4) as recommended by Ishwaran.
min.node.size	Minimal node size. Default is 1.
num.threads	number of threads used for parallel execution. Default is number of CPUs available.
status	status variable, only applicable to survival data. Use 1 for event and 0 for censoring.
save.ranger	Set TRUE if ranger object should be saved. Default is that ranger object is not saved (FALSE).
create.forest	set FALSE if you want to analyze an existing forest. Default is TRUE.
forest	the random forest that should be analyzed if create.forest is set to FALSE. (x and y still have to be given to obtain variable names)
save.memory	Use memory saving (but slower) splitting mode. No effect for survival and GWAS data. Warning: This option slows down the tree growing, use only if you encounter memory problems. (This parameter is transfered to ranger)
case.weights	Weights for sampling of training observations. Observations with larger weights will be selected with higher probability in the bootstrap (or subsampled) samples for the trees.

Value

List with the following components:

• info: list with results from mindep function:

  • depth: mean minimal depth for each variable.

  • selected: variables has been selected (1) or not (0).

  • threshold: the threshold that is used for the selection. (deviates slightly from the original implimentation)

• var: vector of selected variables.

• forest: a list containing: #'

  • trees: list of trees that was created by getTreeranger, addLayer, and addSurrogates functions and that was used for surrogate minimal depth variable importance.

  • allvariables: all variable names of the predictor variables that are present in x.

• ranger: ranger object

References

• Ishwaran, H. et al. (2011) Random survival forests for high-dimensional data. Stat Anal Data Min, 4, 115–132. https://onlinelibrary.wiley.com/doi/abs/10.1002/sam.10103

• Ishwaran, H. et al. (2010) High-Dimensional Variable Selection for Survival Data. J. Am. Stat. Assoc., 105, 205–217. http://www.ccs.miami.edu/~hishwaran/papers/IKGML.JASA.2010.pdf

Examples

# read data
data("SMD_example_data")


# select variables (usually more trees are needed)
set.seed(42)
res = var.select.md(x = SMD_example_data[,2:ncol(SMD_example_data)], y = SMD_example_data[,1], ntree = 10)
res$var

StephanSeifert/SurrogateMinimalDepth documentation built on Aug. 7, 2023, 1:59 a.m.