surv_onestep_complete: One-step TMLE estimator for survival curve (No censoring)

Description Usage Arguments Value

View source: R/1_surv_onestep.R

Description

options to ADD: SL.formula: the covariates to include in SL

Usage

1
2
3
4
surv_onestep_complete(dat, dW, g.SL.Lib = c("SL.glm", "SL.step",
  "SL.glm.interaction"), ht.SL.Lib = c("SL.mean", "SL.glm", "SL.gam",
  "SL.earth"), ..., epsilon.step = 1e-05, max.iter = 1000,
  tol = 1/nrow(dat), T.cutoff = NULL, verbose = TRUE)

Arguments

dat

data.frame with columns T, A, W. All columns with character "W" will be treated as baseline covariates.

dW

binary input vector specifying dynamic treatment (as a function output of W)

g.SL.Lib

SuperLearner library for fitting treatment regression

ht.SL.Lib

SuperLearner library for fitting conditional hazard regression

...

additional options for plotting initial fit curve

epsilon.step

step size for one-step recursion

max.iter

maximal number of recursion for one-step

tol

tolerance for optimization

T.cutoff

manual right censor the data; remove parts dont want to esimate

verbose

to plot the initial fit curve and the objective function value during optimzation

Value

Psi.hat vector of survival curve under intervention

T.uniq vector of time points where Psi.hat gets values (have same length as Psi.hat)

params list of meta-information of estimation

variables list of data summary

initial_fit list of initial fit (hazard, g_1, Delta)


wilsoncai1992/onestep.survival documentation built on Dec. 17, 2017, 12:07 p.m.