R: agrep with vector pattern

Question

I have a vector of patterns, and need to use agrep on them. The problem is that agrep seems to take only one pattern at a time.

patt <- c("test","10 Barrel")
lut  <- c("1 Barrel","10 Barrel Brewing","Harpoon 100 Barrel Series","resr","rest","tesr")

for (i in 1:length(patt)) {
  print(agrep(patt[i],lut,max=1,v=T))
}

Result:

[1] "rest" "tesr"
[1] "1 Barrel"                  "10 Barrel Brewing"         "Harpoon 100 Barrel Series"

for is slow on long patterns, thus attempting to do it in vectorized form:

VecMatch1 = function(string, stringVector){
  stringVector[agrep(string, stringVector, max = 1)]
}
a = VecMatch1(patt,lut)

Warning message:
In agrep(string, stringVector, max = 1) :
  argument 'pattern' has length > 1 and only the first element will be used

May be functions like lapply etc can help? Thanks!!

Serban Tanasa · Accepted Answer · 2015-07-15T16:14:48.653

6

Using lapply:

lapply(patt, agrep, x=lut, max.distance=c(cost=1, all=1), value=TRUE)

[[1]]
[1] "rest" "tesr"

[[2]]
[1] "1 Barrel"                  "10 Barrel Brewing"         "Harpoon 100 Barrel Series"

You can probably get faster performance with dplyr or data.table.

edited Jul 15 '15 at 16:14

answered Jul 15 '15 at 16:12

Serban Tanasa

3,592
2
23
45

Yes, thank you! I tried `lapply` but used wrong syntax - didn't put function in proper place – Alexey Ferapontov Jul 15 '15 at 16:14
Turned out, both methods have comparable speed. – Alexey Ferapontov Jul 15 '15 at 18:49
2

How would you approach it with dplyr or data.table? – Alexey Ferapontov Jul 15 '15 at 20:17

R: agrep with vector pattern

1 Answers1