Extracting numbers from vectors of strings

I have string like this:

years<-c("20 years old", "1 years old") 

I would like to grep only the numeric number from this vector. Expected output is a vector:

c(20, 1) 

How do I go about doing this?

12 Answers

How about

# pattern is by finding a set of numbers in the start and capturing them as.numeric(gsub("([0-9]+).*$", "\\1", years)) 

or

# pattern is to just remove _years_old as.numeric(gsub(" years old", "", years)) 

or

# split by space, get the element in first index as.numeric(sapply(strsplit(years, " "), "[[", 1)) 

Update Since extract_numeric is deprecated, we can use parse_number from readr package.

library(readr) parse_number(years) 

Here is another option with extract_numeric

library(tidyr) extract_numeric(years) #[1] 20 1 

I think that substitution is an indirect way of getting to the solution. If you want to retrieve all the numbers, I recommend gregexpr:

matches <- regmatches(years, gregexpr("[[:digit:]]+", years)) as.numeric(unlist(matches)) 

If you have multiple matches in a string, this will get all of them. If you're only interested in the first match, use regexpr instead of gregexpr and you can skip the unlist.

Here's an alternative to Arun's first solution, with a simpler Perl-like regular expression:

as.numeric(gsub("[^\\d]+", "", years, perl=TRUE)) 

Or simply:

as.numeric(gsub("\\D", "", years)) # [1] 20 1 

A stringr pipelined solution:

library(stringr) years %>% str_match_all("[0-9]+") %>% unlist %>% as.numeric 

We can also use str_extract from stringr

years<-c("20 years old", "1 years old") as.integer(stringr::str_extract(years, "\\d+")) #[1] 20 1 

If there are multiple numbers in the string and we want to extract all of them, we may use str_extract_all which unlike str_extract returns all the macthes.

years<-c("20 years old and 21", "1 years old") stringr::str_extract(years, "\\d+") #[1] "20" "1" stringr::str_extract_all(years, "\\d+") #[[1]] #[1] "20" "21" #[[2]] #[1] "1" 

You could get rid of all the letters too:

as.numeric(gsub("[[:alpha:]]", "", years)) 

Likely this is less generalizable though.

Extract numbers from any string at beginning position.

x <- gregexpr("^[0-9]+", years) # Numbers with any number of digits x2 <- as.numeric(unlist(regmatches(years, x))) 

Extract numbers from any string INDEPENDENT of position.

x <- gregexpr("[0-9]+", years) # Numbers with any number of digits x2 <- as.numeric(unlist(regmatches(years, x))) 

Using the package unglue we can do :

# install.packages("unglue") library(unglue) years<-c("20 years old", "1 years old") unglue_vec(years, "{x} years old", convert = TRUE) #> [1] 20 1 

^{Created on 2019-11-06 by the reprex package (v0.3.0)}

More info:

After the post from Gabor Grothendieck post at the r-help mailing list

years<-c("20 years old", "1 years old") library(gsubfn) pat <- "[-+.e0-9]*\\d" sapply(years, function(x) strapply(x, pat, as.numeric)[[1]]) 

I am interested in this question as it applies to extracting values from the base::summary() function. Another option you might want to consider to extract values from a table is to build a function that takes any entry of your summary() table and transforms it into a useful number. For example if you get:

(s <- summary(dataset)) sv_final_num_beneficiarios sv_pfam_rec sv_area_transf Min. : 1.0 Min. :0.0000036 Min. :0.000004 1st Qu.: 67.5 1st Qu.:0.0286363 1st Qu.:0.010107 Median : 200.0 Median :0.0710803 Median :0.021865 Mean : 454.6 Mean :0.1140274 Mean :0.034802 3rd Qu.: 515.8 3rd Qu.:0.1527177 3rd Qu.:0.044234 Max. :17516.0 Max. :0.8217923 Max. :0.360924 

you might want to extract that 1st Qu for sv_pfam_rec and for that read the 2nd row of the 2nd col. In order to get the formatted single value I made a function

s_extract <- function(summary_entry){ separate(as_tibble(summary_entry), sep = ":", col = value, remove = FALSE, into = c("bad", "good"))[[3]] %>% as.numeric() } 

You just have to feed a summary entry, for example summary_entry = s[3,3] to obtain the Median of sv_area_transf.

It is worth nothing that given that this function is based on separate() it makes it easier to navigate certain cases in which the name of the variable also contains numbers