RStudio Trainer Program

Tidyverse: Split-Apply-Combine - Lecture Note

Tidyverse Korea

1 Lecture Note ¹

• Split-Apply-Combine Strategy
• Brute force programming
• Old fashioned Split-Apply-Combine
• purrr Split-Apply-Combine

2 Brute force programming

library(tidyverse)

iris_tbl <- iris %>% as_tibble()

# calculate `Sepal.Length` mean by Species
species_v <- iris_tbl %>% 
  pull(Species) %>% 
  unique() %>% 
  as.character()

sepal_length_mean <- c()

for(i in seq_along(species_v)) {
  species_tmp <- iris_tbl %>% filter(Species == species_v[i])
  sepal_length_mean[i] <- species_tmp %>% summarise(mean(Sepal.Length)) %>% pull
}

sepal_length_mean

[1] 5.006 5.936 6.588

3 Old fashioned Split-Apply-Combine

# normal function
calc_mean <- function(df) {
  mean_val <- mean(df$`Sepal.Length`)
}

iris_split <- split(iris_tbl, iris_tbl$Species)

lapply(iris_split, calc_mean)

$setosa
[1] 5.006

$versicolor
[1] 5.936

$virginica
[1] 6.588

# lambda function ----
# lapply(iris_split, function(x) mean(x$`Sepal.Length`))

4 purrr Split-Apply-Combine

# dplyr verbs
iris_tbl %>% 
  group_by(Species) %>% 
  summarise(mean_sepal_length = mean(`Sepal.Length`))

# A tibble: 3 x 2
  Species    mean_sepal_length
  <fct>                  <dbl>
1 setosa                  5.01
2 versicolor              5.94
3 virginica               6.59

# group_by, nest, mutate, map pattern
iris_tbl %>% 
  group_by(Species) %>% 
  nest() %>% 
  mutate(mean_sepal_length = map_dbl(data, ~mean(.$`Sepal.Length`)))

# A tibble: 3 x 3
# Groups:   Species [3]
  Species              data mean_sepal_length
  <fct>      <list<df[,4]>>             <dbl>
1 setosa           [50 × 4]              5.01
2 versicolor       [50 × 4]              5.94
3 virginica        [50 × 4]              6.59

---

1. R 병렬 프로그래밍 - “분할-적용-병합(Split-Apply-Combine) 전략”↩