Build.RF.ITR: Builds the random forest of interaction trees

In jinjinzhou/ITR.Tree: Building an ITR Tree

Description

Builds the random forest of interaction trees

Usage

Build.RF.ITR(dat, col.y, col.trt, col.prtx, split.var, ctg = NA, N0 = 20,
  n0 = 5, max.depth = 10, ntree = 500,
  mtry = max(floor(length(split.var)/3), 1), avoid.nul.tree = F)

Arguments

dat	the data set being used to grow the random forest. Required input.
col.y	the response variable. Required input.
col.trt	the treatment indicator. Must be binary. Required input.
col.prtx	the probability of being assigned to treatment group. Required input.
split.var	vector of columns containing the desired splitting variables. Required input.
ctg	identifies the categorical input columns. Defaults to NA. Not available yet.
N0	minimum number of observations needed to call a node terminal. Defaults to 20.
n0	minimum number of treatment/control observations needed in a split to call a node terminal. Defaults to 5.
max.depth	controls the maximum depth of the tree. Defaults to 10.
ntree	sets the number of trees to be generated. Defaults to 500.
mtry	sets the number of randomly selected splitting variables to be included. Defaults to max of length(split.var)/3 rounded down and 1.
avoid.nul.tree	controls if trees with no splits (null trees) are allowed. Defaults to FALSE.

Value

summary of randomly generated trees (summary done by tree)

Examples

forest<-Build.RF.ITR(dat=data, col.y="y", col.trt="trt", col.prtx="prtx", 
                     split.var=3:7)
This builds a forest of 500 trees using the dataset called 'data' with columns
'y', 'trt', and 'prtx' for the outcome, treatement indicator, and probability of being
in treatment group, respectively.  The splitting variables are found in columns 3-7.

jinjinzhou/ITR.Tree documentation built on May 19, 2019, 10:36 a.m.