dplyr_grouping_stuff.md

library(tidyverse)

# Proportion of different types of drives in the data
mpg %>% 
  group_by(drv) %>% 
  summarize(total = n()) %>% 
  # R automatically ungrouped - it'll always ungroup the last group (in this case drv)
  # Percentage - sum(total) is based on the ungrouped 3-row dataset
  mutate(prop = total / sum(total))
#> `summarise()` ungrouping output (override with `.groups` argument)
#> # A tibble: 3 x 3
#>   drv   total  prop
#>   <chr> <int> <dbl>
#> 1 4       103 0.440
#> 2 f       106 0.453
#> 3 r        25 0.107

# Proportion of different types of drives within years
mpg %>% 
  group_by(drv, year) %>% 
  summarize(total = n()) %>% 
  # R automatically ungroups the last layer (year), so now this is grouped by drv
  # If we calculate the proportion now, each drv will sum to 1
  mutate(prop = total / sum(total))
#> `summarise()` regrouping output by 'drv' (override with `.groups` argument)
#> # A tibble: 6 x 4
#> # Groups:   drv [3]
#>   drv    year total  prop
#>   <chr> <int> <int> <dbl>
#> 1 4      1999    49 0.476
#> 2 4      2008    54 0.524
#> 3 f      1999    57 0.538
#> 4 f      2008    49 0.462
#> 5 r      1999    11 0.44 
#> 6 r      2008    14 0.56

# If we explicitly ungroup, we can get the proportion to be across all 6 rows instead of within drv
mpg %>% 
  group_by(drv, year) %>% 
  summarize(total = n()) %>% 
  # Explicitly remove all behind-the-scenes groups and work on the whole dataset
  ungroup() %>% 
  # Now the whole prop column will sum to 1
  mutate(prop = total / sum(total))
#> # A tibble: 6 x 4
#>   drv    year total   prop
#>   <chr> <int> <int>  <dbl>
#> 1 4      1999    49 0.209 
#> 2 4      2008    54 0.231 
#> 3 f      1999    57 0.244 
#> 4 f      2008    49 0.209 
#> 5 r      1999    11 0.0470
#> 6 r      2008    14 0.0598

^{Created on 2020-09-01 by the reprex package (v0.3.0)}