specialFrame: Special frame
In Publish: Format Output of Various Routines in a Suitable Way for Reports and Publication

specialFrame

R Documentation

Special frame

Description

Extract data and design matrix including specials from call

Usage

specialFrame(
  formula,
  data,
  unspecials.design = TRUE,
  specials,
  specials.factor = TRUE,
  specials.design = FALSE,
  strip.specials = TRUE,
  strip.arguments = NULL,
  strip.alias = NULL,
  strip.unspecials = NULL,
  drop.intercept = TRUE,
  response = TRUE,
  na.action = options()$na.action
)

Arguments

`formula`	Formula whose left hand side specifies the event history, i.e., either via Surv() or Hist().
`data`	Data frame in which the formula is interpreted
`unspecials.design`	Passed as is to `model.design`.
`specials`	Character vector of special function names. Usually the body of the special functions is function(x)x but e.g., `strata` from the survival package does treat the values
`specials.factor`	Passed as is to `model.design`.
`specials.design`	Passed as is to `model.design`
`strip.specials`	Passed as `specials` to `strip.terms`
`strip.arguments`	Passed as `arguments` to `strip.terms`
`strip.alias`	Passed as `alias.names` to `strip.terms`
`strip.unspecials`	Passed as `unspecials` to `strip.terms`
`drop.intercept`	Passed as is to `model.design`
`response`	If FALSE do not get response data.
`na.action`	Decide what to do with missing values.

Details

Obtain a list with the data used for event history regression analysis. This function cannot be used directly on the user level but inside a function to prepare data for survival analysis.

Value

A list which contains - the response - the design matrix (see model.design) - one entry for each special (see model.design)

Author(s)

Thomas A. Gerds <tag@biostat.ku.dk>

Examples


## Here are some data with an event time and no competing risks
## and two covariates X1 and X2.
## Suppose we want to declare that variable X1 is treated differently
## than variable X2. For example, X1 could be a cluster variable, or
## X1 should have a proportional effect on the outcome.
d <- data.frame(y=1:7,
                X2=c(2.24,3.22,9.59,4.4,3.54,6.81,5.05),
                X3=c(1,1,1,1,0,0,1),
                X4=c(44.69,37.41,68.54,38.85,35.9,27.02,41.84),
                X1=factor(c("a","b","a","c","c","a","b"),
                    levels=c("c","a","b")))
## define special functions prop and cluster
prop <- function(x)x
cluster <- function(x)x
## We pass a formula and the data
e <- specialFrame(y~prop(X1)+X2+cluster(X3)+X4,
                  data=d,
                  specials=c("prop","cluster"))
## The first element is the response
e$response
## The other elements are the design, i.e., model.matrix for the non-special covariates
e$design
## and a data.frame for the special covariates
e$prop
## The special covariates can be returned as a model.matrix 
e2 <- specialFrame(y~prop(X1)+X2+cluster(X3)+X4,
                   data=d,
                   specials=c("prop","cluster"),
                   specials.design=TRUE)
e2$prop
## and the non-special covariates can be returned as a data.frame
e3 <- specialFrame(y~prop(X1)+X2+cluster(X3)+X4,
                   data=d,
                   specials=c("prop","cluster"),
                   specials.design=TRUE,
                   unspecials.design=FALSE)
e3$design

Publish documentation built on Jan. 18, 2023, 1:08 a.m.