Partial matching when subsetting a dataframe

[Rachelly]

This really struck me.. When sub-setting a data.frame using the [] operator, partial matched names are picked! Is this how it's supposed to work?? I vaguely remember that I discussed this with Renaud once, but don't remember the conclusion we got to.. Thanks! Rachelly.

x=data.frame("A"=c(1,2,3,4), "B"=c(1,2,3,4), "C"=c(1,2,3,4), row.names = c("123","345","1201","22"))

x A B C 123 1 1 1 345 2 2 2 1201 3 3 3 22 4 4 4

x["123",] A B C 123 1 1 1

x["120",] A B C 1201 3 3 3

x["12",] # A row is not found, because there are 2 possible matches A B C NA NA NA NA

x[rownames(x) == "120",] # At least this works!! [1] A B C

<0 rows> (or 0-length row.names)

[renozao]

Yes, this how it is meant to work, although I agree it is a dangerous default behaviour. Check the documentation for data.frame, which recommends using match when exact matching is desired. But you have to be careful with indexes that are in the query and not in the data.frame: look at what x[NA, ] returns:

x[NA, ]
> x[NA, ]
      A  B  C
NA   NA NA NA
NA.1 NA NA NA
NA.2 NA NA NA
NA.3 NA NA NA

No comment on the disastrous potential side effect of this.

Options:

• If your data is actually a matrix (i.e. all columns are of the same atomic type), then you'd rather convert it into a matrix object.
• Alternatively you could use the following trick/function:

x=data.frame("A"=c(1,2,3,4), "B"=c(1,2,3,4), "C"=c(1,2,3,4), row.names = c("123","345","1201","22"))
index <- c('123', '120', '12')
subset_row <- function(x, index){
   res <- x[match(index, rownames(x), nomatch = nrow(x) + 1), ]
   rownames(res) <- index
   res
}
subset_row(x, index)