dmlmt: Double machine learning for binary and multiple treatments'

Description Usage Arguments Value

View source: R/dmlmt.R

Description

This function estimates treatment effects for binary and multiple treatments using Double Machine Learning.

Usage

1
2
3
dmlmt(x, t, y, family = "gaussian", pl = TRUE, cs = TRUE, q = 1,
  cl = NULL, print = FALSE, se_rule = NULL, w = FALSE,
  parallel = FALSE, ...)

Arguments

x

Matrix of covariates (N x p matrix)

t

Vector of treament indicators. Will be ordered from 0 to T-1.

y

Vector of outcomes

family

Outcome type. Default is "gaussian". For binary outcomes choose "binomial".

pl

If TRUE Post-Lasso is used to estimate nuisance parameters, if FALSE Lasso

cs

If TRUE, common support will be checked

q

Quantile used for enforcing common support

cl

Vector with cluster variables can be provided

print

If TRUE, supporting information is printed

se_rule

If not NULL, define, e.g., c(-1,1) to get 1SE and 1SE+ rule

w

If TRUE, implied weights are calculated (only if pl=TRUE)

parallel

If TRUE, cross-validation of post_lasso_cv parallelized

...

Pass glmnet and post_lasso_cv options

Value

dmlmt returns the results of the estimated average treatment effects and the potential outcomes. If specified, results of different SE rules and implied weights are returned.


MCKnaus/dmlmt documentation built on Dec. 4, 2020, 9:48 a.m.