What is the best way to tidy data and extract numbers (double) from data frame in R?

Question

I have a data frame that contains survey responses. What is the best way to extract the numbers and change them to a double type variable?

Here is a little sample:

a <- ("10.5", "about 30", "25 per month")
tibble(a)

I have tried

parse_double(a)

and it seems like I am close. Any help is appreciated

Does this answer your question? [Extracting decimal numbers from a string](https://stackoverflow.com/questions/19252663/extracting-decimal-numbers-from-a-string) — camille, Oct 15 '21 at 21:21
Also 4 more options [in this post](https://stackoverflow.com/q/28819761/5325862) and another 11 options [here](https://stackoverflow.com/q/14543627/5325862) that just need to be amended to include the decimal point — camille, Oct 15 '21 at 21:25

score 1 · Answer 1 · answered Oct 15 '21 at 19:12

1

We need parse_number

library(readr)
parse_number(a)
[1] 10.5 30.0 25.0

The difference is that parse_double works on character vectors with only digits + . as the characters whereas parse_number extracts the numeric part from a string which also include non-numeric characters

data

a <- c("10.5", "about 30", "25 per month")

answered Oct 15 '21 at 19:12

akrun

874,273
37
540
662

My solution is faster ;))) – manro Oct 15 '21 at 21:07
Did you have any tasks in your practice, where were some requirements to the execution time? – manro Oct 15 '21 at 21:10
If we want to achieve a time efficency of the code - we must use basic functions of R? – manro Oct 15 '21 at 21:24

score 1 · Answer 2 · edited Oct 16 '21 at 22:11

I know a solution too (from base package)

a <- c("10.5", "about 30", "25 per month")
as.numeric(gsub("[[:alpha:]]", "", a)) 


 > as.numeric(gsub("[[:alpha:]]", "", a))
 [1] 10.5 30.0 25.0
 > end_time <- Sys.time()
 > end_time - start_time
 Time difference of 0.01400113 secs
 > start_time <- Sys.time()
 > parse_number(a)
 [1] 10.5 30.0 25.0
 > end_time <- Sys.time()
 > end_time - start_time
 Time difference of 0.1500092 secs

My solution is faster than that provided by Akrun.

What is the best way to tidy data and extract numbers (double) from data frame in R?

2 Answers2

data