tstmle3: Context-specific causal effect of single-time point...

In podTockom/tstmle: Data-adaptive Targeted Maximum Likelihood Estimation of Causal Effects of Interventions on Single Time Series

Context-specific causal effect of single-time point intervention

Description

This function estimates the causal effect of a single time-point intervention on the outcome within the same time point. Here, time is defined as observing a single intervention and outcome, and possibly multiple covariates. Intervention is imposed within each unit of time.

Usage

tstmle3(
  data,
  node_list,
  parameter = "ATE",
  learner_list,
  Co = 5,
  cvtmle = FALSE,
  fold_fn = "folds_rolling_origin",
  first_window = 100,
  validation_size = 50,
  gap = 0,
  batch = 50,
  window_size = 100
)

Arguments

data	data.frame object containing the data.
node_list	node list reflecting the relationships between variables.
parameter	target parameter for targeting. Default is the average over time Context-Specific Average Treatment Effect.
learner_list	learner list containing the learners used for the conditional expectation of outcome and propensity score.
Co	user-specified Markov order for the fixed dimensional summary measure.
cvtmle	default is FALSE.
fold_fn	cross-validation scheme, as defined by origami. See ?origami::fold_funs for detailed explanations. For time-series, implemented cross-validation schemes are folds_rolling_origin and folds_rolling_window.
first_window	first window size used for training. Only relevant if the set cross-validation is folds_rolling_origin.
validation_size	number of time points used for validation.
gap	number of time points between training and validation set.
batch	number of time points added in the next next cross-validation fold.
window_size	window size used for training. Only relevant if the set cross-validation is folds_rolling_window.

Value

An object of class tstmle.

tmlePsi

Average treatment effect estimated using TMLE.

iptwPsi

Average treatment effect estimated using IPTW.

tmleSE

Standard error for the TMLE estimated parameter.

tmleSD

Standard deviation for the TMLE estimated parameter.

tmleCI

Confidence Interval for the TMLE estimated parameter.

IC

Influence function.

steps

Number of steps until convergence of the iterative TMLE.

initialData

Initial estimates of g, Q.

tmleData

Final updates estimates of g, Q and clever covariates.

tmle_fit

The full tmle3 fit object.

podTockom/tstmle documentation built on Aug. 11, 2022, 12:46 a.m.