How do I create new column data based on regex match in R-CodePudding

I have some tweet author location data that I'm looking to reclassify to country. For example, taking a vector of United States 'states' I want to check (regex) for a match and add a "United States" entry to the country column.

Example data:

states = c("CA", "OH", "FL", "TX", "MN") # all the states
tweets$location = data.frame("my bed", "Minneapolis, MN", "Paris, France", "Los Angeles, CA")

What i've tried:

# This seems to do the matching part well
filter(str_detect(location, paste(usa_data$Code, collapse = "|")))

# nested for loop
for (i in length(tweets$location)){
  for (state in states){
    if (grepl(state, tweets$location[i])){
      tweets$country[i] = "USA"
      break
    }
  }
}

Desired output (based on example input):

tweets$country = data.frame(NA, "USA", NA, "USA")

I'm relatively new to R, therefore any help will be greatly appreciated.

CodePudding user response：

We can use grepl along with ifelse for a base R solution:

states = c("CA", "OH", "FL", "TX", "MN") # all the states
tweets$location = data.frame("my bed", "Minneapolis, MN", "Paris, France", "Los Angeles, CA")
regex <- paste0("\\b(?:", paste(states, collapse="|"), ")\\b")
tweets$country <- ifelse(grepl(regex, tweets$location), "USA", NA)

CodePudding user response：

If you prefer a dplyr solution, but very similar to Tim's answer

library(dplyr)
states <- c("CA", "OH", "FL", "TX", "MN") # all the states


tweets <- tibble(location = c(
  "my bed", "Minneapolis, MN", "Paris, France",
  "Los Angeles, CA"
))

tweets %>%
  mutate(country = if_else(stringr::str_detect(
   string =  location,
   pattern = paste0(
      "\\b(?:", paste(states,
        collapse = "|"
      ),
      ")\\b"
    )
  ),
  "United States", "NA"
  ))