rsfit: Estimation and inference for relative contrast function under...

In muxuanliang/RSICF: Estimation and Inference for the Relative Contrast Function under a Monotonic Single-index Model Assumption

Description

Estimation and inference for relative contrast function under a monotonic single-index model assumption

Usage

rsfit(
  covariate,
  response,
  treatment,
  splitIndex = NULL,
  propensityModel = "glmnet",
  estimatedPropensity = NULL,
  outcomeModel = "kernel",
  estimatedOutcome = NULL,
  lossType = "logistic",
  weights = NULL,
  tol = 0.001,
  propensityFormula = NULL,
  outcomeFormula = NULL,
  parallel = FALSE,
  constraint = TRUE,
  boundaryPoint = c(-1, 1),
  efficient = TRUE,
  local = TRUE
)

Arguments

covariate	input matrix of dimension nobs x nvars. Each raw is a observation, each column is a covariate
response	numeric response
treatment	is a vector of binary value representing two treatment, 0 or 1.
propensityModel	Similar to outcomeModel.
estimatedPropensity	estimated propensity score p(trt=1|X). This should be used only when the propensity is estimated by a parametric model.
outcomeModel	this selects method to estimate the outcome when estimatedOutcome=NULL. Options include lm, glmnet, kernel, and gam. If lm is used, user also need to input the outcomeFormula like y~x used in lm. By default, kernel regression is selected.
estimatedOutcome	estimated outcome model E(y|trt, X). This should be used only when the outcome is estimated by a parametric model. It should be a list (list(control=..., treatment=...)).
boundaryPoint	please specify appropriate range for the single-index. Default is [-1,1].

Value

A list

betaAN

The one-step updated coefficients estimates

sigmaAN

The estimated sd of the estimated coefficients

betaAN.renorm

The one-step updated coefficients estimates after re-normalized such that \|beta\|_2=1

sigmaAN.renorm

The estimated sd of the normalized estimates

fit

A list used for predict.rsfit which predicts the relative contrast

Author(s)

Muxuan Liang

References

Muxuan Liang, Menggang Yu (2022). Relative Contrast Estimation and Inference for Treatment Recommendation.

muxuanliang/RSICF documentation built on Feb. 1, 2022, 12:30 a.m.