如何使用列名列表进行group_by和汇总？

Question

Basically, I want to loop through, group the data by the columns in "list.group", and then create summary statistics for every column in "list.avg", "list.max", and "list.min" so that the columns are mpg_avg, wt_avg, hp_avg, mpg_max, hp_max... mpg_min, hp_min, etc.

data("mtcars")
    list.avg <- list("mpg","wt","hp")
    list.max <- list("mpg","hp","wt","qsec")
    list.min <- list("mpg","hp","wt","qsec")
    list.group <- list("cyl","vs","am","gear","carb")

So I should have a separate table for each column in list.group.

Answer 1

First, it's helpful to have all the avg/max/min variables in a single list.

to_summarise <- 
  list(mean = c("mpg","wt","hp"),
       max = c("mpg","hp","wt","qsec"),
       min = c("mpg","hp","wt","qsec"))

Now we can map over list.group , and within each list.group value, imap over to_summarise and then merge all the results together.

library(tidyverse)

map(list.group, ~{
  grouped <- 
    mtcars %>% 
      group_by_at(.x) 
  out <- 
    imap(to_summarise, ~{
            grouped %>% 
              summarise_at(.x, setNames(list(get(.y)), .y))
    })
  out %>% 
    reduce(merge, by = .x)
})

Output

#     [[1]]
#   cyl mpg_mean  wt_mean   hp_mean mpg_max hp_max wt_max qsec_max mpg_min hp_min wt_min
# 1   4 26.66364 2.285727  82.63636    33.9    113  3.190    22.90    21.4     52  1.513
# 2   6 19.74286 3.117143 122.28571    21.4    175  3.460    20.22    17.8    105  2.620
# 3   8 15.10000 3.999214 209.21429    19.2    335  5.424    18.00    10.4    150  3.170
#   qsec_min
# 1     16.7
# 2     15.5
# 3     14.5
# 
# [[2]]
#   vs mpg_mean  wt_mean   hp_mean mpg_max hp_max wt_max qsec_max mpg_min hp_min wt_min
# 1  0 16.61667 3.688556 189.72222    26.0    335  5.424     18.0    10.4     91  2.140
# 2  1 24.55714 2.611286  91.35714    33.9    123  3.460     22.9    17.8     52  1.513
#   qsec_min
# 1     14.5
# 2     16.9
# 
# [[3]]
#   am mpg_mean  wt_mean  hp_mean mpg_max hp_max wt_max qsec_max mpg_min hp_min wt_min
# 1  0 17.14737 3.768895 160.2632    24.4    245  5.424     22.9    10.4     62  2.465
# 2  1 24.39231 2.411000 126.8462    33.9    335  3.570     19.9    15.0     52  1.513
#   qsec_min
# 1    15.41
# 2    14.50
# 
# [[4]]
#   gear mpg_mean  wt_mean  hp_mean mpg_max hp_max wt_max qsec_max mpg_min hp_min wt_min
# 1    3 16.10667 3.892600 176.1333    21.5    245  5.424    20.22    10.4     97  2.465
# 2    4 24.53333 2.616667  89.5000    33.9    123  3.440    22.90    17.8     52  1.615
# 3    5 21.38000 2.632600 195.6000    30.4    335  3.570    16.90    15.0     91  1.513
#   qsec_min
# 1    15.41
# 2    16.46
# 3    14.50
# 
# [[5]]
#   carb mpg_mean wt_mean hp_mean mpg_max hp_max wt_max qsec_max mpg_min hp_min wt_min
# 1    1 25.34286  2.4900    86.0    33.9    110  3.460    20.22    18.1     65  1.835
# 2    2 22.40000  2.8628   117.2    30.4    175  3.845    22.90    15.2     52  1.513
# 3    3 16.30000  3.8600   180.0    17.3    180  4.070    18.00    15.2    180  3.730
# 4    4 15.79000  3.8974   187.0    21.0    264  5.424    18.90    10.4    110  2.620
# 5    6 19.70000  2.7700   175.0    19.7    175  2.770    15.50    19.7    175  2.770
# 6    8 15.00000  3.5700   335.0    15.0    335  3.570    14.60    15.0    335  3.570
#   qsec_min
# 1    18.61
# 2    16.70
# 3    17.40
# 4    14.50
# 5    15.50
# 6    14.60

Answer 2

The 'avg' is not a function in R . Instead, it can be mean . So, changing the object identifier name from list.avg to list.mean , keep the list. objects into a list , then loop through the named list with imap , remove the prefix list. with str_remove , using group_by_at group by the common grouping elements, then summarise_at the values that we loop while applying the function we get from the prefix removed names on those columns

library(tidyverse)
list.mean <- list("mpg","wt","hp")
lst(list.mean, list.max, list.min) %>% 
   imap(~ {

   func <- str_remove(.y, '^list\\.')
    vars1 <- unlist(.x)



  mtcars %>%
     group_by_at(unlist(list.group)) %>%
      summarise_at(vars(vars1), ~ get(func)(.))


  })

Answer 3

Use map to loop through list.group , use group_by_at to group at each element of list.group as they are strings then summarize at the required columns and finally binds all together.

library(purrr)
library(dplyr)
map(list.group, ~mtcars %>% 
          #.x will be "cyl", "vs" ... etc 
          group_by_at(.x) %>% 
          {bind_cols(summarise_at(.,unlist(list.avg), list(avg=mean)),
                     summarise_at(.,unlist(list.min), list(min=min)),
                     summarise_at(.,unlist(list.max), list(max=max))
                     )
          }
    )