average_partial_effect: Estimate average partial effect via augmented balancing

In swager/amlinear: Augmented Minmimax Linear Estimation

Estimate average partial effect via augmented balancing

Description

Estimate average partial effect via augmented balancing

Usage

average_partial_effect(X, Y, W, balance.method = c("minimax", "plugin"),
  zeta = 0.5, fitted.model = NULL, alpha = 1, standardize = TRUE,
  solver = c("ECOS", "SCS"), verbose = TRUE)

Arguments

X	the input features
Y	the observed response (real valued)
W	the effect variable (real valued)
balance.method	how the balancing weights gamma are derived
zeta	tuning parameter for selecting approximately balancing weights
fitted.model	optional pre-fitted regression adjustment
alpha	tuning paramter for glmnet
standardize	whether non-binary features should be noramlized
solver	convex optimzer used by CVXR for minimax weights
verbose	whether the optimizer should print progress information

Value

ATE estimate with standard error estimate. Also returns “linear” point estimate of the form sum gamma_i Yi, as in Donoho (1994), for comparison.

swager/amlinear documentation built on Aug. 30, 2023, 4:21 a.m.