compute_asam: Average Standardized Absolute Mean Distance

In ygeunkim/propensityml: Propensity score weighting using machine learning methods

Description

computes average standardized absolute mean distance (ASAM)

Usage

compute_asam(
  data,
  treatment,
  trt_indicator = 1,
  outcome,
  exclude = NULL,
  object = NULL,
  formula = NULL,
  method = c("logit", "rf", "cart", "SVM"),
  weighting = c("IPW", "SIPW"),
  mc_col = NULL,
  sc_col = NULL,
  parallel = FALSE,
  ...
)

Arguments

data	data
treatment	column name of treatment
trt_indicator	value that indicates the unit is treated, e.g. 1 or TRUE
outcome	outcome variable included in the data. It should be specified because it is not covariate.
exclude	Additional columns to exlude
object	A propmod object if already fitted.
formula	If not, write a formula to be fitted. Remember that you don't have to worry about group variable. .SD do exclude by.
method	Estimating methods "logit" - ps_glm "rf" - ps_rf "cart" - ps_cart "SVM" - ps_svm
weighting	Weighting methods, IPW or SIPW
mc_col	Indicator column name for MC simulation if exists
sc_col	Indicator column name for various scenarios if exists
parallel	parallelize some operation
...	Additional arguments of fitting functions

Details

For each covariate, compute absolute (standardized difference of means between treatment and control groups), and take average. Denote that standardization is done by sd of treatment group covariates.

Lower ASAM means that treatment and control groups are more similar w.r.t. the given covariates.

References

Lee, B. K., Lessler, J., & Stuart, E. A. (2010). Improving propensity score weighting using machine learning. Statistics in Medicine. Statistics in Medicine, 29(3), 337-346. https://doi.org/10.1002/sim.3782

ygeunkim/propensityml documentation built on Jan. 1, 2021, 1:44 p.m.