R remove multiple text strings in data.table

Question

I have a vector of words to remove from a data.table DT as follows.

wordstoremove <- c("Simpson", "Flander", "Nahasapeemapetilon", "Spuckler", "Wiggum")

DT <- structure(list(vid = c("Simpsons", "Flanders", "Nahasapeemapetilons", 
                             "Spucklers", "Wiggums"), wr1 = c("Homer Simpson", "Ned Flanders", 
                                                              "Apu Nahasapeemapetilon", "Cletus Spuckler", "Chief Wiggum"), 
                     wr2 = c("Bart Simpson", "Rod Flanders", "Manjula Nahasapeemapetilon", 
                             "Brandine Spuckler", "Ralph Wiggum"), wr3 = c("Marge Simpson", 
                                                                           "Todd Flanders", "Sanjay Nahasapeemapetilon", NA, "Sarah Wiggum"
                             )), .Names = c("vid", "wr1", "wr2", "wr3"), row.names = c(NA, 
                                                                                       -5L), class = c("data.table", "data.frame"))

DT
                   vid                    wr2                        wr2                       wr3
1:            Simpsons          Homer Simpson               Bart Simpson             Marge Simpson
2:            Flanders           Ned Flanders               Rod Flanders             Todd Flanders
3: Nahasapeemapetilons Apu Nahasapeemapetilon Manjula Nahasapeemapetilon Sanjay Nahasapeemapetilon
4:           Spucklers        Cletus Spuckler          Brandine Spuckler                        NA
5:             Wiggums           Chief Wiggum               Ralph Wiggum              Sarah Wiggum

I know I can use the solution in R remove multiple text strings in data frame.

How to do this using data.table to minimise copying of data?

Andriy T. · Accepted Answer · 2015-07-16T06:43:31.797

6

Try this:

library(data.table)
foo <- function(x) gsub(paste0(wordstoremove, collapse="s?|"), "", x)
DT[, names(DT)[-1] := lapply(.SD, foo), .SDcols = names(DT)[-1]]
DT
#                    vid     wr1       wr2     wr3
# 1:            Simpsons  Homer      Bart   Marge 
# 2:            Flanders    Ned       Rod    Todd 
# 3: Nahasapeemapetilons    Apu   Manjula  Sanjay 
# 4:           Spucklers Cletus  Brandine       NA
# 5:             Wiggums  Chief     Ralph   Sarah

edited Jul 16 '15 at 06:43

answered Jul 14 '15 at 07:40

Andriy T.

2,020
12
23

3

usually just `names(DT)` instead of `colnames(DT)` is used – MichaelChirico Jul 14 '15 at 17:46
2

and maybe use `s?` to allow for the possibility of plurals – MichaelChirico Jul 14 '15 at 17:47
Yes, I missed it after Crop's edit. What do you mean by `s`? – Andriy T. Jul 15 '15 at 06:50
1

I'd think a for-loop with `set()` would be even more efficient (would be nice to check on meaningful data sizes, e.g., 2GB data). – Arun Jul 15 '15 at 07:38
@Arun, I cant't find out the way to do it fast, but seems that it is. See update – Andriy T. Jul 15 '15 at 09:38
`?` will match the previous character 0 or 1 times – MichaelChirico Jul 15 '15 at 11:34
@MichaelChirico, you mean `wordstoremove <- c("Simpsons?", "Flanders?", ...)` ? – Andriy T. Jul 15 '15 at 13:06
nono, I mean to add that to the regex so it matches "Simpson" AND "Simpsons" – MichaelChirico Jul 15 '15 at 13:33
1

@MichaelChirico, yeah, you're right. `paste` will do it for me. See update – Andriy T. Jul 16 '15 at 06:41

R remove multiple text strings in data.table

1 Answers1