group_map and multiple arguments

user_2602640 0
Viewed 29

I have a large dataset that is grouped. For each row within the group, I need to extract a vector from one of the columns, and determine its minimum. The vector's length would be determined by the value in another column. I can do this with a combination of pmap_dbl() and group_map(), but I was wondering if there's a cleaner way of doing this while staying in the dplyr universe.

A toy example:

library(dplyr)
library(purrr)

dat <- data.frame(temp = rnorm(10, 10, 2), 
              start = c(1:5, 1:5), 
              end = c(2, 2, 4, 5, 5, 2, 3, 4, 5, 5),
              group = c(rep("a", 5), rep("b", 5)))

Function to get the vectorized needed output, this returns the min value of the vector defined by start/end for each row:

fun.try <- function(dat, ...) {
    pmap_dbl(list(start = dat$start, end = dat$end), function(x, start, end) 
        min(x$temp[start:end]), x = dat)                  
                              }

Applying the function by group:

dat %>%
   group_by(group) %>%
   group_map(.f = fun.try)

Is there a cleaner way of doing this, preferably using an anonymous function? Thank you!

purrr dplyr r
1 Answers

We can use map2 as there are only variables

library(dplyr)
library(purrr)
dat %>% 
 group_by(group) %>%
 mutate(Minval = map2_dbl(start, end, ~ min(temp[.x:.y]))) %>% 
 ungroup
user_37322710
Related