MartingalePlot

Produces a plot of the martingale or deviance residuals, smoothed with lowess and a confidence band based on visreg asymptotics. Additionally, a bootstrap confidence band can be obtained.

MartingalePlot(DFR, TIMES, STATUS, METRIC, COLUMNS = NULL,
  TYPE = "martingale", SPANLOESS = 2/3, NBOOT = 0, XLIM = NULL,
  YLIM = NULL, XLAB = NULL, YLAB = "Martingale residuals",
  MAIN = NULL, SUB = NULL, PCH = 1, PCEX = 1)

Arguments

DFR	data.frame containing the time to event and metric variable
TIMES	name of variable in DFR: EITHER Time to event variable of class "Surv" OR time component of time to event
STATUS	name of variable in DFR: status component von time to event (needed only if TIMES is not of class "Surv")
METRIC	name of metric variable in DFR to be investigated
COLUMNS	= NULL, vector of further covariate names in DFR != METRIC to adjust for in calculating the residuals.
TYPE	="martingale", specifies type of residuum, "martingale" or the more symmetric "deviance"
SPANLOESS	=2/3, smoothing parameter for lowess smoother
NBOOT	=0, number of Bootstrap simulations. Set to > 0 (e.g. 1000), if additional bootstrap confidence band desired
XLIM	=NULL, set range on x-axis manually instead of automatically
YLIM	=NULL, set range on y-axis manually instead of automatically
XLAB	=NULL, set x-axis label manually instead of automatically
YLAB	="Martingale residuals", standard y-axis label
MAIN	=NULL, for no main title set MAIN=""
SUB	=NULL, subtitle below XLAB, if unspecified adjusted for: COLUMNS
PCH	=1 pointtype for martingalresidues
PCEX	=1 pont size (cex)

Value

produces a plot

Details

MartingalePlot

Examples

if (FALSE) {
require(MASS)
data(VA)
# The Martingal plot shows the functional form of the dependence of 
#              a time to event and a metric variable.
MartingalePlot(VA,"stime","status","age")
MartingalePlot(VA,"stime","status","age",SPANLOESS = .9)
MartingalePlot(VA,"stime","status","age", XLIM=c(40,70))

MartingalePlot(VA,"stime","status","Karn")
MartingalePlot(VA,"stime","status","Karn",SPANLOESS = .9)

# Martingal residues of age after adjusting linearly for karno
MartingalePlot(VA,"stime","status","age",COLUMNS ="Karn",
            SPANLOESS = .9, XLIM=c(40,70))
}