rcDT: Constructs an rcDT model

In kdoub5ha/rcITR: Risk Controlled ITR Discovery

Description

Constructs a risk controlled decision tree (rcDT) given an efficacy and risk outcome.

Usage

rcDT(data, split.var, test = NULL, ctg = NULL, efficacy = "y",
  risk = "r", col.trt = "trt", col.prtx = "prtx", risk.control = TRUE,
  risk.threshold = NA, lambda = 0, min.ndsz = 20, n0 = 5,
  stabilize = TRUE, stabilize.type = c("linear", "rf"),
  use.other.nodes = TRUE, mtry = length(split.var), max.depth = 15,
  AIPWE = FALSE, extremeRandomized = FALSE, print.summary = TRUE)

Arguments

data	data.frame. Data used to construct rcDT model. Must contain efficacy variable (y), risk variable (r), binary treatment indicator coded as 0 / 1 (trt), propensity score (prtx), candidate splitting covariates (split.var).
split.var	numeric vector. Columns of spliting variables.
test	data.frame of testing observations. Should be formatted the same as 'data'.
ctg	numeric vector corresponding to the categorical input columns. Defaults to NULL. Not available yet.
efficacy	char. Efficacy outcome column. Assumes larger values are desirable Defaults to 'y'.
risk	char. Risk outcome column. Assumes smaller values are desirable Defaults to 'r'.
col.trt	char. Treatment indicator column name. Should be of form 0/1 or -1/+1.
col.prtx	char. Propensity score column name.
risk.control	logical. Should risk be controlled? Defaults to TRUE.
risk.threshold	numeric. Desired level of risk control.
lambda	numeric. Penalty parameter for risk scores. Defaults to 0, i.e. no constraint. Optional arguments
min.ndsz	numeric specifying minimum number of observations required to call a node terminal. Defaults to 20.
n0	numeric specifying minimum number of treatment/control observations needed in a split to declare a node terminal. Defaults to 5.
stabilize	logical indicating if efficacy should be modeled using residuals. Defaults to TRUE.
stabilize.type	character specifying method used for estimating residuals. Current options are 'linear' for linear model (default) and 'rf' for random forest.
use.other.nodes	logical. Should global estimator of objective function be used. Defaults to TRUE.
mtry	numeric specifying the number of randomly selected splitting variables to be included. Defaults to number of splitting variables.
max.depth	numeric specifying maximum depth of the tree. Defaults to 15 levels.
AIPWE	logical. Should AIPWE (TRUE) or IPWE (FALSE) be used. Not available yet.
extremeRandomized	logical. Experimental for randomly selecting cutpoints in a random forest model. Defaults to FALSE and users should change this at their own peril.
print.summary	logical. Should a summary of the tree building be printed? Defaults to TRUE for single trees.

Value

Summary of rcDT model

tree	data.frame with the following: Each 'node' begins with "0" indicating the root node, followed by a "1" or "2" indicating the less than (or left) child node or greater than (or right) child node. Additionally, the number of observations 'size', number treated 'n.1', number on control 'n.0', and treatment effect 'trt.effect' summaries are provided. The splitting information includes the column of the chosen splitting variable ‘var', the variable name ’vname', the direction the treatment is sent 'cut.1' ("r" for right child node, and "l" for left), the chosen split value 'cut.2', and the estimated value function 'score'.
y	efficacy values used in modeling. Will likely differ from original input 'y' if stabilization was used
risk.threshold	value of risk control used
data	input dataset
fit.y	fitted model for residuals is 'stabilize' was used
split.var	splitting covariates used

Examples

set.seed(123)
dat <- generateData()
# Generates tree using simualated EMR data with splitting variables located in columns 1-4.
tree <- rcDT(data = dat, 
             split.var = 1:10, 
             risk.threshold = 2.75, 
             lambda = 1)

kdoub5ha/rcITR documentation built on Aug. 5, 2020, 9:05 p.m.