R: Abbreviate state names in strings

Question

I have strings with state names in them. How do I efficiently abbreviate them? I am aware of state.abb[grep("New York", state.name)] but this works only if "New York" is the whole string. I have, for example, "Walmart, New York". Thanks in advance!

Let's assume this input:

x = c("Walmart, New York", "Hobby Lobby (California)", "Sold in Sears in Illinois")

Edit: desired outputs will be a la "Walmart, NY", "Hobby Lobby (CA)", "Sold in Sears in IL". As you can see from here, state can appear in many ways in a string

What are your expectations? To get "Walmart, N.Y."? But what are the assumptions? — Wiktor Stribiżew, Oct 02 '15 at 18:50
"Walmart, NY" as per conventions (and that is what the `state.abb` would have done if it worked with full string) — Alexey Ferapontov, Oct 02 '15 at 18:51
My suggestion: Split your string into "first part" (just the city?) and "state part". Make a table mapping from full state names to abbreviations. Merge. Don't combine the two parts of the string again, because why would you? — Frank, Oct 02 '15 at 18:54
@Frank, good working suggestion. Altho not a very efficient, as apriori I do not know how exactly the state name will appear in the string. But something to start with — Alexey Ferapontov, Oct 02 '15 at 18:56
Seems like you'll have to use an `*apply()` construct , probably with `Map()` since it's one-to-one. Something like `Map(sub, state.name, state.abb, x)` where `x` is the vector you wish to change — Rich Scriven, Oct 02 '15 at 19:03
@RichardScriven, Yes! Altho I am horrible with any `apply` family functions. Still need to learn the syntax. How would you furnish it? — Alexey Ferapontov, Oct 02 '15 at 19:06
Refresh for the edited comment, I think that should work. Well, not perfectly but it's a start — Rich Scriven, Oct 02 '15 at 19:06

Josh O'Brien · Accepted Answer · 2015-10-02T19:50:23.643

Here's a base R way, using gregexpr(), regmatches(), and regmatches<-(), :

abbreviateStateNames <- function(x) {
    pat <- paste(state.name, collapse="|")
    m <- gregexpr(pat, x)
    ff <- function(x) state.abb[match(x, state.name)]
    regmatches(x, m) <- lapply(regmatches(x, m), ff)
    x
}

x <- c("Hobby Lobby (California)", 
       "Hello New York City, here I come (from Greensboro North Carolina)!")

abbreviateStateNames(x)
# [1] "Hobby Lobby (CA)"                                
# [2] "Hello NY City, here I come (from Greensboro NC)!"

Alternatively -- and quite a bit more naturally -- you can accomplish the same thing using the gsubfn package:

library(gsubfn)

pat <- paste(state.name, collapse="|")
gsubfn(pat, function(x) state.abb[match(x, state.name)], x)
[1] "Hobby Lobby (CA)"                                
[2] "Hello NY City, here I come (from Greensboro NC)!"

You don't see `regmatches<-` every day. Nice one – Rich Scriven Oct 02 '15 at 19:19 — Rich Scriven, Oct 02 '15 at 19:19

R: Abbreviate state names in strings

1 Answers1