Home / Questions / 328342
Accepted
Alejandro Carrera
Asked: 2020-02-11 16:55:37 +0800 CST 2020-02-11 16:55:37 +0800 CST

Is it possible to optimize this string extraction function using regex in R?

  • 772

Suppose we have a string of last names like the following:

nom <- c("Perez Conchito", "Juanin Juanharry", "Von Bola")

I have designed a small function to extract each part of the string and then gather them into a data.frame with two columns.

extract_apellidos <- function(x) {
    first_split <- strsplit(x, " ")
    split_un <- unlist(first_split)
    primer_ap <- split_un[c(TRUE, FALSE)] #Nos da el primer apellido
    segundo_ap <- split_un[c(FALSE,TRUE)] #Nos da el segundo apellido
    data.frame(Primer_ap=primer_ap, Segundo_ap=segundo_ap)
}

extract_apellidos(nom)


  Primer_ap Segundo_ap
1     Perez   Conchito
2    Juanin  Juanharry
3       Von       Bola

As you can see, it works correctly. However, I would like to know if it is possible to optimize it using regular functions since I suspect that this would allow me to reduce the number of steps used. I thank you in advance for any guidance on this.

r

1 Answers

  1. Best Answer

    Alejandro, at least with base R, whatever you do with regular expressions, you're going to end up with a list, that is, in the same place it leaves you:

    first_split <- strsplit(x, " ")

    For example with:

    regmatches(nom, regexec("(.*)\\s(.*)", nom))

    we have obtained a list, with vectors of 3 elements, the complete string that made "match" and the first and second words. You can't gain practically anything from your routine (check, just as you have an error eto_todos, it doesn't exist, I understand it should be split_un). On the other hand, using regular expressions to separate words from a space is unnecessarily complex.

    Another thing is, if you use stringr, since you can take advantage str_match()of capturing groups and that their output is already an array, with this you can shorten the code a bit:

    library("stringr")

match_df <- as.data.frame(str_match(nom, "(.*)\\s(.*)"))[, c(2,3)]
colnames(match_df) <- c("Primer_ap", "Segundo_ap")

    But it also has an extra advantage: by respecting the methodology of tydverse, the return is consistent with the input object, so that if it does not find a pattern, it will still return a row for that case:

    nom <- c("Perez Conchito", "Juanin Juanharry", "Von Bola", "Pedro")
match_df <- as.data.frame(str_match(nom, "(.*)\\s(.*)"))[, c(2,3)]
colnames(match_df) <- c("Primer_ap", "Segundo_ap")
match_df

  Primer_ap Segundo_ap
1     Perez   Conchito
2    Juanin  Juanharry
3       Von       Bola
4      <NA>       <NA>
    • 2