tstmle01: Time-Series Targeted Minimum Loss Estimation for Binary Time...

In podTockom/tstmle01: Targeted Maximum Likelihood Estimation of Causal Effects of Interventions on a Single Binary Time Series

Description

This function returns the final estimate of the expected value the outcome at time t, under intervention (or no intervention) as specified by the user. In addition, it returns the variance of the estimate, confidence intervals, and p-value as dictated by specified alpha.

Usage

tstmle01(data, freqY, freqA, freqW, t, Anode, intervention1,
  intervention2 = NULL, MC = 100, B = 100, N = 100, maxIter = 50,
  tol = 10^-5, alpha = 0.05, param = "ate")

Arguments

data	data.frame object containg the time series with relevant time ordering.
freqY	A numeric specifying the Markov order for Y nodes.
freqA	A numeric specifying the Markov order for A nodes.
freqW	A numeric specifying the Markov order for W nodes.
t	Outcome time point of interest. It must be greater than the intervention node A.
Anode	Intervention node.
intervention1	Specify g^*, of P(A|past). Right now supports only 1/0 type interventions.
intervention2	Specify g^* to compate to. Right now supports only 1/0 type interventions.
MC	How many Monte Carlo samples should be generated.
B	How many samples to draw from P, as part of the h-density estimation.
N	How many sample to draw from P^*, as part of the h-density estimation.
maxIter	Maximum number of iterations.
tol	Lower bound for epsilon.
alpha	alpha
param	Return average treatment effect (ate) or risk ratio (rr).

Value

An object of class tstmle01.

psi

Estimate of the target parameter.

var.psi

Variance, based on the influence function.

CI

Confidence intervals.

IC	Influence curve.

podTockom/tstmle01 documentation built on May 14, 2019, 2:03 a.m.