CODEWITHSUNDEEP
rdplyrtidyverse
Eric Fail
Eric Fail

Reputation: 7928

Simple way to rotate dplyr's tibble summary

I am looking for a simple way to rotate 's tibble summary.

Say I am doing something like this,

# install.packages(c("dplyr"), dependencies = TRUE)
library(dplyr) 

mtcars %>% 
    group_by(am)  %>%
    summarise(
        n = n(),
        Mean_disp = mean(disp),
        Mean_hp = mean(hp),
        Mean_qsec = mean(qsec),
        Mean_drat = mean(drat)
             )
#>     # A tibble: 2 x 6
#>      am     n Mean_disp  Mean_hp Mean_qsec Mean_drat
#>   <dbl> <int>     <dbl>    <dbl>     <dbl>     <dbl>
#> 1     0    19  290.3789 160.2632  18.18316  3.286316
#> 2     1    13  143.5308 126.8462  17.36000  4.050000

But, what I would like is to get an output more or less like this,

#> # A tibble: 5 x 2
#>    am <dbl>             0           1
#>   n <int>              19          13
#>   Mean_disp <dbl>     290.3789    143.5308
#>   Mean_hp <dbl>       160.2631    126.8462
#>   Mean_qsec <dbl>      18.183158   17.36000
#>   Mean_drat <dbl>       3.286316    4.050000

I realize I can use t(), but that transform the tibble to a list and mess up the formatting.

Upvotes: 2

Views: 2218

Answers (2)

zx8754
zx8754

Reputation: 56219

Maybe gather then spread again:

library(dplyr) 
library(tidyr)

mtcars %>% 
  group_by(am)  %>%
  summarise(
    n = n(),
    Mean_disp = mean(disp),
    Mean_hp = mean(hp),
    Mean_qsec = mean(qsec),
    Mean_drat = mean(drat)) %>%
  gather(key = key, value = value, -am) %>% 
  spread(key = am, value = value)


# # A tibble: 5 x 3
#         key        `0`      `1`
# *     <chr>      <dbl>    <dbl>
# 1 Mean_disp 290.378947 143.5308
# 2 Mean_drat   3.286316   4.0500
# 3   Mean_hp 160.263158 126.8462
# 4 Mean_qsec  18.183158  17.3600
# 5         n  19.000000  13.0000

Another option, gather before group_by, then get mean for all selected columns, then spread again (but not sure how to add n()):

mtcars %>% 
  select(am, disp, hp, qsec, drat) %>% 
  gather(key = key, value = value, -am) %>% 
  group_by(am, key)  %>%
  summarise(myMean = mean(value)) %>%
  spread(key = am, value = myMean)

# # A tibble: 4 x 3
#     key        `0`      `1`
# * <chr>      <dbl>    <dbl>
# 1  disp 290.378947 143.5308
# 2  drat   3.286316   4.0500
# 3    hp 160.263158 126.8462
# 4  qsec  18.183158  17.3600

Upvotes: 6

icj
icj

Reputation: 749

  new_tibble <- as.data.frame(t(mt_cars_df)) %>% 
  as_tibble()

  new_tibble$name <- names(mt_cars_df) %>%
    select(name, V1, V2)

according to this, row.names are deprecated in tibbles, so adding them as a key column would the logical way of handling the situation. This leaves a tibble that requires reordering the columns.

Upvotes: 0

Related Questions