CODEWITHSUNDEEP
rdplyr
Jess A.
Jess A.

Reputation: 95

Replacing original variable using parse_number and mutate in dplyr

I have some data that contains a set of variables that have a mix of numbers and characters, such as "1 - Completely Disagree." I would like to remove the characters and keep only the numbers, but retain the variables within the original dataset under their original name. So, I have a variable called "systematic1" with the value "1 - Completely Disagree. I would like it to become just "1" in numeric form within the original dataset. This what I have:

systematic1 systematic2 systematic3
1 - Completely Disagree 7 - Completely Agree 7 - Completely Disagree
5 - Somewhat Agree 4 - Neither Agree nor Disagree 6 - Agree

This is the desired output:

systematic1 systematic2 systematic3
1 7 7
5 4 6

I've been able to mostly accomplish this using the following code:

data %>%
   select(systematic1:withdrawn) %>% #select range 
   select_if(is.character) %>% #keep only character vars
   mutate_all(~parse_number(.,na=c("Not sure"))) #parse out number, treating "Not sure" as NA

But that produces a new dataframe with those transformed variables. I would like to keep the variables in the original dataset, only transformed.

Upvotes: 1

Views: 580

Answers (1)

Martin Gal
Martin Gal

Reputation: 16978

You could use

library(dplyr)
library(stringr)

data %>% 
  mutate(
    across(
      systematic1:withdrawn & where(is.character), 
      ~ifelse(str_detect(.x, "\\d+"), str_extract(.x, "\\d+"), "Not sure")
      )
    )

which returns

# A tibble: 2 x 3
  systematic1 systematic2 systematic3
  <chr>       <chr>       <chr>      
1 1           7           7          
2 5           4           6

Upvotes: 1

Related Questions