stable: Main function to calculate stability coefficients

In dtree: Decision Trees

Description

Main function to calculate stability coefficients

Usage

stable(formula, data, methods = c("lm", "rpart", "tree", "ctree", "evtree"),
  samp.method = "repeatedcv", tuneLength = 3, n.rep = 100,
  bump.rep = 50, parallel = FALSE, ncore = detectCores() - 1,
  roundVal = 1, stablelearner = FALSE, subset = FALSE, perc.sub = 0.75,
  weights = NULL)

Arguments

formula	a formula, weight a response to left of ~.
data	Data frame to run models on
methods	Which tree methods to use. Defaults: lm, rpart, tree, ctree, evtree. Also can use "rf" for random forests
samp.method	Sampling method. Refer to caret package trainControl() documentation. Default is repeated cross-validation. Other options include "cv" and "boot".
tuneLength	Number of tuning parameters to try. Applies to train()
n.rep	Number of times to replicate each method
bump.rep	Number of repetitions for bumping
parallel	Whether to run all reps in parallel
ncore	Number of cores to use
roundVal	How much to round cut points when calculating stability
stablelearner	Whether or not to use the stablelearner package to calculate stability
subset	Whether to subset
perc.sub	What fraction of data to put into train dataset. 1-frac.sub is allocated to test dataset. Defaults to 0.75
weights	Optional weights for each case.

Examples

## Not run: 
library(MASS) # for boston data
data(Boston)
stab.out <- stable(formula=medv ~.,data=Boston,
            methods=c("rpart"),samp.method="cv",
            tuneLength=2, n.rep=5, parallel=TRUE)
stab.out

## End(Not run)

dtree documentation built on May 1, 2019, 9:19 p.m.