将函数应用于2个字符向量的返回列表-Purrr

Question

I'm doing some work using purrr and hoping for a complete piped solution for this problem. I am using sapply but think this isn't the optimal solution. It works for this small demo, but in the real data ch1 is length >50,000 and ch2 is >100.

library(stringr)
library(purrr)

ch1 <- c("something very interesting or perhaps it is not", "lions, tigers and elephants are safari animals", "once upon a time there was a big castle",
         "I have not seen anything as a big as elephants")

ch2 <- c("big", "not")

For each element of ch2 we want to see if they occur in each element of ch1 .

str_detect(ch1, ch2[1]) # FALSE FALSE  TRUE  TRUE
str_detect(ch1, ch2[2]) # TRUE FALSE FALSE  TRUE

Trying to use purrr to apply function over all of ch1 :

ch1 %>% map_lgl(str_detect(., ch2[2]))  # TRUE FALSE FALSE  TRUE

I can do this for the entirety of ch2 using sapply :

sapply(ch2, function(x) ch1 %>% map_lgl(str_detect(., x)))

      big   not
[1,] FALSE  TRUE
[2,] FALSE FALSE
[3,]  TRUE FALSE
[4,]  TRUE  TRUE

However, with the real dataset I think there must be a full purrr solution - something like using map2 ie working on two lists - but obviously it can't be that particular one as it requires lists of equal lengths.

Answer 1

The following, which would probably be a bit faster on large data sets than the code in your post, returns a list of vectors.

library(stringr)
library(purrr
lst <- ch2 %>% split(ch2) %>% 
  map( ~ str_detect(ch1, .x))

To return a matrix, you could use the following:

mat <- ch2 %>% split(ch2) %>% 
      map( ~ str_detect(ch1, .x)) %>%
      map_call(cbind)

However, since map_call is just a thin wrapper for do.call , it may be a bit slow. If you could use dplyr and work with a data.frame as the result, the following may be a little faster:

library(dplyr)
df <- ch2 %>% split(ch2) %>% 
      map( ~ str_detect(ch1, .x)) %>%
      as_data_frame() 

Added

The following is a solution which produces a matrix with named columns using map2

# solution using map2
mat2 <- ch1 %>% list %>%
        map2(ch2,  ~ str_detect(.x, .y)) %>%
        map_call(cbind)
colnames(mat2) <- ch2

Perhaps the most straightforward which produces a matrix with column names is:

names(ch2) <- ch2
mat3 <- ch2 %>% map( ~ str_detect(ch1, .x)) %>% 
        map_call(cbind)