Delete multiple rows based on some constrains - r

Question

I'm using R and I am trying to delete some rows from a data frame based on some constrains. so, if I got
dat <- data.frame(Cs=c("c1","c2","c3","c4","c5","c6"),
R1=sample(c("Y","N"),6,replace=TRUE), R2=sample(c("Y","N"),6,replace=TRUE),
R3=sample(c("Y","N"),6,replace=TRUE), R4=sample(c("Y","N"),6,replace=TRUE),
R5=sample(c("Y","N"),6,replace=TRUE), R6=sample(c("Y","N"),6,replace=TRUE))
I'd like to delete all the rows having a "N" at some given columns such as R1, R3, R4. For one single column, I found this solution: delete row for certain constrains
d <- dat[dat[,"R1"]!="N",]
which works fine. but if I put multiple columns as
d <- dat[dat[,c("R1","R3","R4")]!="N",]
I got lots of extra rows full of NA. So where am I wrong?

Answer 1

You can use
dat[rowSums(dat[, c("R1","R3","R4")] == "N") == 0, , drop=FALSE]
# Cs R1 R2 R3 R4 R5 R6
#5 c5 Y Y Y Y Y Y
Or, if you don't like excessive typing:
dat[!rowSums(dat[c('R1','R3','R4')]=='N'),]
This will first test each "cell" of columns "R1", "R3" and "R4" of your data whether it is equal to "N" and then calculate the sums of TRUE values per row. If no "N" is present in a row, the sum is equal to 0 and will be kept. I added drop=FALSE to keep the structure as a data.frame.
Note after a comment by OP:
If you subset only 1 column of a data.frame without specifying a drop=TRUE option, the default behavior of [.data.frame is to coerce the resulting 1-column-data.frame to an atomic vector. Then, rowSums wouldn't work on that resulting vector. To avoid that, change your code to:
dat[!rowSums(dat[,'R1', drop=FALSE]=='N'), ]
Sample data:
set.seed(5)
dat <- data.frame(Cs=c("c1","c2","c3","c4","c5","c6"),
R1=sample(c("Y","N"),6,replace=TRUE), R2=sample(c("Y","N"),6,replace=TRUE),
R3=sample(c("Y","N"),6,replace=TRUE), R4=sample(c("Y","N"),6,replace=TRUE),
R5=sample(c("Y","N"),6,replace=TRUE), R6=sample(c("Y","N"),6,replace=TRUE))

Answer 2

You could make a 'keep'-variable consisting of booleans for each row:
keep <- apply(dat[,c("R1","R3","R4")],
MARGIN=1,
FUN=function(x){all(x!='N')})
res <- dat[keep,]
> res
Cs R1 R2 R3 R4 R5 R6
1 c1 Y Y Y Y Y Y
data:
seed used: 1234
dat <- structure(list(Cs = structure(1:6, .Label = c("c1", "c2", "c3",
"c4", "c5", "c6"), class = "factor"), R1 = structure(c(2L, 1L,
1L, 1L, 1L, 1L), .Label = c("N", "Y"), class = "factor"), R2 = structure(c(2L,
2L, 1L, 1L, 1L, 1L), .Label = c("N", "Y"), class = "factor"),
R3 = structure(c(2L, 1L, 2L, 1L, 2L, 2L), .Label = c("N",
"Y"), class = "factor"), R4 = structure(c(1L, 1L, 1L, 1L,
1L, 1L), .Label = "Y", class = "factor"), R5 = structure(c(2L,
1L, 1L, 1L, 1L, 2L), .Label = c("N", "Y"), class = "factor"),
R6 = structure(c(2L, 2L, 2L, 1L, 2L, 1L), .Label = c("N",
"Y"), class = "factor")), .Names = c("Cs", "R1", "R2", "R3",
"R4", "R5", "R6"), row.names = c(NA, -6L), class = "data.frame")