randomForestSRC (version 2.8.0)

holdoutvimp: Hold out variable importance (VIMP)

Description

Hold out VIMP is calculated from the error rate for trees grown with and without a variable. Applies to all families.

Usage

# S3 method for rfsrc
holdoutvimp(formula, data,
  ntree = 1000 * ncol(data) / vtry,
  ntree.max = 2000,
  nsplit = 10,
  ntime = 50,
  mtry = NULL,
  vtry = 1,
  fast = FALSE,
  verbose = TRUE,
  ...)

Arguments

formula

A symbolic description of the model to be fit.

data

Data frame containing the y-outcome and x-variables.

ntree

Number of trees used for growing the forest.

ntree.max

Maximum number of trees used when calculating prediction error for determing hold out VIMP.

nsplit

Non-negative integer value specifying number of random split points used to split a node (deterministic splitting corresponds to the value zero and is much slower).

ntime

Integer value used for survival to constrain ensemble calculations to a grid of ntime time points.

mtry

Number of variables randomly selected as candidates for splitting a node.

vtry

Number of variables randomly selected to be held out when growning a tree.

fast

Use fast random forests, rfsrcFast, in place of rfsrc? Improves speed but is less accurate.

verbose

Provide verbose output?

...

Further arguments to be passed to rfsrc.

Value

Hold out VIMP for each variable. For multivariate forests, hold out VIMP is calculated for each of the target outcomes.

Details

Prior to growing a tree, a random set of vtry features are held out. Tree growing proceeds as usual with the remaining features. Once the forest is grown, hold out VIMP for a given variable v is calculated as follows. Gather all trees where v was held out and calculate OOB prediction error. Next gather all trees were v was not held out and calculate OOB prediction error. Hold out VIMP for v is the difference between these two values. Thus hold out VIMP measures the importance of a variable when that variable is truly removed from tree growing.

Accuracy of hold out VIMP depends heavily on the size of the forest. If the number of trees is too small, then number of trees where v is held out will be small, and the resulting OOB error will have high variance. Thus, ntree should be set fairly high - we recommend using 1000 times the number of features. Increasing vtry is another way to increase number of hold out trees. In particular, number of trees needed should decrease linearly with vtry. Keep in mind however that intrepretation of holdout VIMP is altered when vtry is different than 1. This is likely to be more of a concern in low dimensional settings.

Uses the new get.tree option in predict to extract specific trees from a forest and the hidden option vtry in rfsrc. The latter creates a hidden array holdout.array of zeroes and ones indicating which variable to hold out in a tree where number of rows equals number of features and number of columns equals number of trees. The array can also be passed as a hidden option but is not checked for coherence so users should be careful when doing so.

References

Lu M. and Ishwaran H. (2018). Expert Opinion: A prediction-based alternative to p-values in regression models. J. Thoracic and Cardiovascular Surgery, 155(3), 1130--1136.

See Also

vimp

Examples

# NOT RUN {
## ------------------------------------------------------------
## Boston housing example
## ------------------------------------------------------------

if (library("mlbench", logical.return = TRUE)) {

  data(BostonHousing)
  hv <- holdoutvimp(medv ~ ., BostonHousing)
  print(hv)

}

## ------------------------------------------------------------
## Multivariate regression analysis
## ------------------------------------------------------------

hv <- holdoutvimp(cbind(mpg, cyl) ~., mtcars)
print(hv)

## ------------------------------------------------------------
## White wine classification example
## ------------------------------------------------------------

data(wine, package = "randomForestSRC")
wine$quality <- factor(wine$quality)
hv <- holdoutvimp(quality ~ ., wine, vtry = 5)
print(100 * hv)


## ------------------------------------------------------------
## pbc survival example
## ------------------------------------------------------------

data(pbc, package = "randomForestSRC")
hv <- holdoutvimp(Surv(days, status) ~ ., pbc, splitrule = "random")
print(100 * hv)

## ------------------------------------------------------------
## WIHS competing risk example
## ------------------------------------------------------------

data(wihs, package = "randomForestSRC")
hv <- holdoutvimp(Surv(time, status) ~ ., wihs, ntree = 1000)
print(100 * hv)

# }