R help - Dec 2005 - Delete missing values

I am trying to delete rows containing missing values from a groupeddata object.
Several of the columns are character (sexChar, HAPI, rs2304785) the rest are
numeric. For some reason I am excluding all rows with missing values. Your
suggestions for corrections would be appreciated.

This did not work
	GC2 <- GC[c("logtg" != NA & "ctime" != NA &
!is.na("sexChar") & !is.na("HAPI") &
"logfirsttg" != NA & "BMI" != NA & !is.na(GC$
		rs2304795)),  ] 
nor did
	GC2 <- GC["logtg" != NA & "ctime" != NA &
!is.na("sexChar") & !is.na("HAPI") &
"logfirsttg" != NA & "BMI" != NA & !is.na(GC$
		rs2304795),  ] 

John

John Sorkin M.D., Ph.D.
Chief, Biostatistics and Informatics
Baltimore VA Medical Center GRECC and
University of Maryland School of Medicine Claude Pepper OAIC

University of Maryland School of Medicine
Division of Gerontology
Baltimore VA Medical Center
10 North Greene Street
GRECC (BT/18/GR)
Baltimore, MD 21201-1524

410-605-7119 
- NOTE NEW EMAIL ADDRESS:
jsorkin at grecc.umaryland.edu

On Wed, 2005-12-14 at 21:34 -0500, John Sorkin wrote:
> I am trying to delete rows containing missing values from a
> groupeddata object. Several of the columns are character (sexChar,
> HAPI, rs2304785) the rest are numeric. For some reason I am excluding
> all rows with missing values. Your suggestions for corrections would
> be appreciated.
> 
> This did not work
> 	GC2 <- GC[c("logtg" != NA & "ctime" != NA &
!is.na("sexChar") & !
> is.na("HAPI") & "logfirsttg" != NA &
"BMI" != NA & !is.na(GC$
> 		rs2304795)),  ] 
> nor did
> 	GC2 <- GC["logtg" != NA & "ctime" != NA &
!is.na("sexChar") & !
> is.na("HAPI") & "logfirsttg" != NA &
"BMI" != NA & !is.na(GC$
> 		rs2304795),  ] 
> 
> John
John,

You cannot use:

  Values != NA

and get the TRUE/FALSE results of the boolean comparison of Values that
are not equal to NA.

For example:
> a <- sample(c(NA, 1:5), 20, replace = TRUE)
> a
 [1]  2  3  3  1  3  4  5  3 NA  4  3  2  1  2  2 NA  2  2 NA  1
> a != NA
 [1] NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA

or 
> a == NA
 [1] NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA


NA is undefined, so by definition, any comparisons to NA, as above, will
be as well.  Simply put:
> NA == NA
[1] NA      # Note that this is not TRUE


That is why there is a specific function to be used, which you have in
some cases above. That is is.na().
> !is.na(a)
 [1]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE
[12]  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE FALSE  TRUE


which then can be used as such:
> a[!is.na(a)]
 [1] 2 3 3 1 3 4 5 3 4 3 2 1 2 2 2 2 1


In the case of a data frame (which a groupedData object contains), you
can use complete.cases() to access the rows that do not have missing
values.  So, if your initial object is called GC, you should be able to
use:

  GC2 <- GC[complete.cases(GC), ]

An alternative is to use na.omit() as follows:

  GC2 <- na.omit(GC)

See ?complete.cases and ?na.omit for more information.

HTH,

Marc Schwartz