cdgd1_ml: Perform conditional decomposition via machine learning

In cdgd: Causal Decomposition of Group Disparities

Perform conditional decomposition via machine learning

Description

Perform conditional decomposition via machine learning

Usage

cdgd1_ml(
  Y,
  D,
  G,
  X,
  Q,
  data,
  algorithm,
  alpha = 0.05,
  trim1 = 0,
  trim2 = 0,
  weight = NULL
)

Arguments

Y	Outcome. The name of a numeric variable (can be binary and take values of 0 and 1).
D	Treatment status. The name of a binary numeric variable taking values of 0 and 1.
G	Advantaged group membership. The name of a binary numeric variable taking values of 0 and 1.
X	Confounders. A vector of variables names.
Q	Conditional set. A vector of names of numeric variables.
data	A data frame.
algorithm	The ML algorithm for modelling. "nnet" for neural network, "ranger" for random forests, "gbm" for generalized boosted models.
alpha	1-alpha confidence interval.
trim1	Threshold for trimming the propensity score. When trim1=a, individuals with propensity scores lower than a or higher than 1-a will be dropped.
trim2	Threshold for trimming the G given Q predictions. When trim2=a, individuals with G given Q predictions lower than a or higher than 1-a will be dropped.
weight	Sampling weights. The name of a numeric variable. If unspecified, equal weights are used. Technically, the weight should be a deterministic function of X only (note that this is different from the unconditional decomposition).

Value

A dataframe of estimates.

Examples

# This example will take a minute to run.

data(exp_data)

set.seed(1)

results <- cdgd1_ml(
Y="outcome",
D="treatment",
G="group_a",
X="confounder",
Q="Q",
data=exp_data,
algorithm="gbm")

results

cdgd documentation built on June 16, 2025, 9:06 a.m.