On Feb 19, 2010, at 1:36 AM, K. Elo wrote:
Dear Anna,
19.02.2010 08:17, Anna Carter wrote:
(1) If the dataset contains some variables having all the entries = 0
and while analysing I want to delete those pericular columns, how do
acheive this. i.e.
Let's suppose 'df' is your data frame, then:
subset(df, select=which(colSums(df)!=0))
should do the work :)
It would not work if there were paired negative and positive values or
any collection that summed to zero.
> dataset1 <-
structure(list(sr_no = 1:4, var1 = c(5L, 3L, 4L, 11L), var2 = c(0,
0, 1, -1), var3 = c(3L, 2L, 4L, 1L), var4 = c(1L, 9L, 7L, 6L),
var5 = c(0L, 0L, 0L, 0L)), .Names = c("sr_no", "var1", "var2",
"var3", "var4", "var5"), row.names = c(NA, -4L), class = "data.frame")
Perhaps:
> idx <- vector()
> for (x in seq_along(names(dataset1))) if (all(dataset1[, x] == 0))
{ } else{ idx<- c(idx, x)}
> dataset1[, idx]
sr_no var1 var2 var3 var4
1 1 5 0 3 1
2 2 3 0 2 9
3 3 4 1 4 7
4 4 11 -1 1 6
Or a modification to Kimmo Elo's code which would still "break" if any
columns were character:
> subset(dataset1, select=which(colSums(abs(dataset1))!=0))
sr_no var1 var2 var3 var4
1 1 5 0 3 1
2 2 3 0 2 9
3 3 4 1 4 7
4 4 11 -1 1 6
HTH,
Kimmo
--
David Winsemius, MD
Heritage Laboratories
West Hartford, CT
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
