GLM.IPW: IPW estimates for a time-specific GLM using the new weight
In michellezhou2009/NCCIPW: NCCIPW

View source: R/GLM.IPW.R

GLM.IPW

R Documentation

IPW estimates for a time-specific GLM using the new weight

Description

IPW estimates of model parameters and accuracy parameters for a time-specific generalized linear model with perturbed counterparts.

Usage

GLM.IPW(
  formula,
  data,
  family = binomial(link = logit),
  id,
  case,
  control,
  m0,
  t0,
  weight.type = "HT",
  yes.match = FALSE,
  control.matching = list(Mdat = NULL, aM = NULL),
  control.accuracy = list(type = "FPR", u0 = 0.05),
  yes.ptb = FALSE,
  control.ptb = list(seed = 123, n.ptb = 500, CaseID = NULL)
)

Arguments

`formula`	a formula object, with the response on the left of a `~` operator, and the terms on the right. The response must be a survival object as returned by the `Surv` function.
`data`	a `data.frame` in which to interpret the variables named in the formula.
`family`	a description of the error distribution and link function to be used in the generalized linear model, which is the argument used in the `glm` function. The default is `binominal(link=logit)`.
`id`	character, which is the variable name in the `data` corresponding to the column that identifies the subjects.
`case`	character, which is the variable name in the `data` corresponding to the column that indicates whether each subject is a sub-cohort case. The value is 1 if the subject is a sub-cohort case, and it is 0 if the subject is not a sub-cohort case.
`control`	character, which is the variable name in the `data` corresponding to the column that indicates whether each subject is a sub-cohort control. The value is 1 if the subject is a sub-cohort control, and it is 0 if the subject is not a sub-cohort control. #' @param m0 a numerical value which is the number of controls selected for each case.
`t0`	a numerical value, which is the prediction time period, used for creating the binary outcome status: =1 if the event time is smaller than `t0`, = 0 otherwise.
`weight.type`	character which indicate which sampling weight to be used and whose value could be "sam" the modified Samuelsen's weight by assigning the inverse selection probability to event cases; "new" the modified Samuelsen's weight by assigning 0 to event cases; "HT" the Horvitz-Thompson type of weight, which is the default value.
`yes.match`	logical indicating whether the controls are selected with or without matching; the default value is `FALSE`.
`control.matching`	a list which includes two elements: `Mdat` and `aM`. `Mdat` is a `data.frame` which includes the matching variables, and its default value is `NULL` if the selection is without matching. `aM` is a numerical vector which includes the matching constraints for each matching variable, and its default value is `NULL` if the selection is without matching. The length of 'aM' should be the same as the number of columns of `Mdat`.
`control.accuracy`	a list which includes the arguments used in estimating the accuracy measures. It includes two elements: `type` and `u0`. `type` is a character which could be "FPR" or "TPR", and the default value is "FPR". `u0` is a numerical value for the accuracy measure specified in `type`, and the default value is 0.05.
`yes.ptb`	logical indicating if obtaining the perturbed counterparts of the IPW estimates, and the default value is `FALSE`.
`control.ptb`	a list of elements that control the perturbation procedure. `seed` is a numerical value used in the function `set.seed`, and its default value is 123. `n.ptb` is a numerical value indicating the number of perturbed counterparts generated, and its default value is 100. `CaseID` is character, which is the variable name in the `data` corresponding to the column that listed the case that each subject that is selected in the sub-cohort for.

Value

a list of two elements: estimates and ptb. estimates is a list of IPW estimates of the model parameters (coef) and accuracy parameters (acc). If yes.ptb=TRUE, ptb is a list of the perturbed counterparts for the IPW estimates of the model parameters (coef) and accuracy parameters (acc); otherwise, ptb = NULL.

Examples

data("myexample")
GLM.IPW(formula=Surv(time,status)~marker1+marker2,
      data=myexample$data,
      id="id",
      case="case",
      control="control",
      m0=3,t0=1,
      weight.type = "HT",
      yes.match=T,control.matching=list(Mdat=myexample$Mdat,aM=myexample$aM),
      yes.ptb=TRUE,control.ptb=list(n.ptb=10,CaseID="CaseID"))

michellezhou2009/NCCIPW documentation built on April 25, 2022, 3:39 p.m.