fit_fluctuation: Fit One-Dimensional Fluctuation Model for Updating Initial...
In txshift: Efficient Estimation of the Causal Effects of Stochastic Interventions

Description Usage Arguments Details Value

Fit One-Dimensional Fluctuation Model for Updating Initial Estimates

fit_fluctuation(
  Y,
  Qn_scaled,
  Hn,
  ipc_weights = rep(1, length(Y)),
  method = c("standard", "weighted"),
  flucmod_tol = 50
)

`Y`	A `numeric` vector corresponding to an outcome variable.
`Qn_scaled`	An object providing the value of the outcome evaluate after inducing a shift in the exposure. This object should be passed in after being constructed by a call to `est_Q`.
`Hn`	An object providing values of the auxiliary ("clever") covariate, constructed from the treatment mechanism and required for targeted minimum loss estimation. This object object should be passed in after being constructed by a call to `est_Hn`.
`ipc_weights`	A `numeric` vector that gives inverse probability of censoring weights for each observation. These are generated by invoking the routines for estimating the censoring mechanism.
`method`	A `character` giving the type of regression to be used in traversing the fluctuation sub-model. The available choices are "weighted" and "standard". Consult the literature for details on the differences.
`flucmod_tol`	A `numeric` indicating the largest value to be tolerated in the fluctuation model for the targeted minimum loss estimator.

Procedure for fitting a one-dimensional fluctuation model to update the initial estimates of the outcome regression based on the auxiliary covariate. These updated estimates are subsequently used to construct the TML estimator of the counterfactual mean under a modified treatment policy.

A list containing the fluctuation model (a glm object) produced by logistic regression, a character vector indicating the type of fluctuation (whether the auxiliary covariates was used as a weight or included directly in the model formula), the updated estimates of the outcome regression under the shifted value of the exposure, and the updated estimates of the outcome regression under the natural value of exposure.

txshift documentation built on Feb. 11, 2022, 1:08 a.m.