R/split_data.R

In flexrsurv: Flexible Relative Survival Analysis

split.data <- function(jeudata, bands, entry, exit, fail, include=names(jeudata), name.runningtime=".t")    
{
	
	
	# Split data in jeudata with breacks in parameter bands
	#----------------------------------
	
	# entry is 1e-12 to prevent further problems
	
	entry <- ifelse(entry <  1e-12, 1e-12, entry )
	Ljeudata <- Lexis( entry = list(.entry = entry), 
			exit = list( .entry = exit ),
			exit.status = fail,
			data= jeudata[, include],
			merge=TRUE, notes=FALSE)
#   without entry (v1.1.2)                  
#   Ljeudata <- Lexis(   exit = list( .entry=exit ),
#                  exit.status = fail,
#                  data=jeudata[, include],
#                  merge=TRUE )
	splitjeudata <- splitLexis( Ljeudata, breaks = bands, time.scale=".entry")
	
	names(splitjeudata) <- ifelse(names(splitjeudata)=="lex.Xst", ".fail", names(splitjeudata))
#   names(splitjeudata) <- ifelse(names(splitjeudata)=="lex.id",  ".expand", names(splitjeudata))
	names(splitjeudata) <- ifelse(names(splitjeudata)=="lex.dur",  "tik", names(splitjeudata))
	splitjeudata$.exit <- splitjeudata$.entry + splitjeudata$tik
	
	splitjeudata <- splitjeudata[, c(1:2, dim(splitjeudata)[2], 3:(dim(splitjeudata)[2]-1))]
	
	
	splitjeudata$tik <- splitjeudata$.exit-splitjeudata$.entry 
	
	splitjeudata$orate <- splitjeudata$rate
	splitjeudata$orate1 <- ifelse(splitjeudata$.fail==0, 0, splitjeudata$rate)
	splitjeudata$newrate<- rep(0, dim(splitjeudata)[1])
	splitjeudata$newrate[cumsum(table(splitjeudata$lex.id))] <- splitjeudata$orate[cumsum(table(splitjeudata$lex.id))]
	splitjeudata$rate <- splitjeudata$newrate
	
	
	# if .fail == 0, the running time (intnum) is the middle of exit and entry time
	#-------------------------------------------------
	splitjeudata$intnum<- ifelse(splitjeudata$.fail==1,
			splitjeudata$.exit,
			(splitjeudata$.entry+splitjeudata$.exit)/2)
	
	# rename the runnig time name (intnum vs name.runningtime)
	names(splitjeudata)<- ifelse(names(splitjeudata)=="intnum", name.runningtime, names(splitjeudata))
	
	return(splitjeudata)  
}