[18884] in s-news-athena
[S] "runs" of missing data in a vector
daemon@ATHENA.MIT.EDU (Barker, Chris {PHAR~Palo Alto})
Fri Aug 6 18:56:39 1999
Content-Return: allowed
Date: Fri, 06 Aug 1999 15:51:13 -0700
From: "Barker, Chris {PHAR~Palo Alto}" <CHRIS.BARKER@ROCHE.COM>
To: "'s-news@wubios.wustl.edu'" <s-news@wubios.wustl.edu>
Message-Id: <B9E51B9719BDD111AB640000F80197CD0164635C@rplmsem2.pal.roche.com>
Mime-Version: 1.0
Content-Type: text/plain; CHARSET=US-ASCII
(My apology in advance for the message length - I attached a short function)
I have a large dataframe with 600 patients, each patient measured daily for
21 days.
I want to identify patterns of missing data so I can prepare groupedData
plots according to pattern of missing data.
I developed a small function (below) which helps me identify
patterns or "runs" of missing data. I use tapply and the
function returns a list with an entry for each patient with
a frequency table of the "runs".
I'd like to convert this list to a data frame which holds
the patient number and the frequency of each "run" for that patient.
(This issue has been discussed in recent posts on this news server, but I
haven't had luck in using the methods discussed, e.g. "expand.grid")
Any suggestions appreciated.
-------function follows----------------------
To clarify, in test vector below, there are 3 "runs" of NA, two that last a
day and a third for 2 days.
test <- c(11,NA,12,NA,18,NA,NA)
I invoke the function using tapply
pt.run.count <- tapply( response.var ,patient.number,FUN=missing.data.runs)
(I'd like to convert pt.run.count to a dataframe and retain the patient
number)
missing.data.runs <- function(vector.test){
zz<-NA;runs<-0 ; runs.vector <-c(rep(NA,length(vector.test)))
if ( !is.na(max(test,na.rm=T) ) ) {
for (i in 1:(length(vector.test)))
{
if (i==1){runs<-0}
if (!is.na(vector.test[i])) {
runs<-0
runs.vector[i] <-NA
}
if ( is.na(vector.test[i]) ) {
runs<-runs+1
runs.vector[i]<-runs
if ( i>1) {if (is.na(vector.test[i-1])){ runs.vector[i-1] <-NA }} }
}
}
else runs.vector <- NA
if ( !is.na(max(runs.vector,na.rm=T)) ) { zz <- table(runs.vector) }
else zz<-c(length(vector.test))
if ( max( !is.na(vector.test) )==1 ) {runs.vector }
if ( !is.na(max(vector.test)) ) { zz <- 0 }
return(zz)
}
Chris Barker, Ph.D.
Principal Pharmacoeconomic Statistician
Roche Pharma Business - Palo Alto
(650)-852-3152
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
