dat <- data.frame(gender=c('M','M','M','M','M','F','F','F','F','F'),
                  num=c(1,2,3,1,2,3,1,2,3,1),
                  h=c(170,180,190,180,170,150,160,170,160,150),
                  w=c(80,70,100,80,60,50,50,60,60,50))
dat$BMI <- dat$w/(dat$h/100)^2

table(dat$gender, dat$num)

##    
##     1 2 3
##   F 2 1 2
##   M 2 2 1

책의 113쪽을 보자.

여기서 aggregate이나 by와 같은 함수를 자세히 설명해 놓지 않았다.

왜냐하면 이 함수들은 dplyr 또는 data.table 패키지를 사용해서도 동일한 결과를 산출할 수 있기 때문이다. 그리고 요즘에는 잘 사용하지 않는 함수들이다. 흔히 말하는 legacy code라고 말할 수 있다.

하지만 그 의미를 이해할 필요도 있으므로 dplyr로 번역을 해보았다.

tapply(dat$h, list(dat$gender, dat$num), mean)
require(dplyr); require(tidyr)
dat %>% group_by(gender, num) %>% summarise(mean = mean(h)) %>% spread(key='num', value='mean')

## `summarise()` regrouping output by 'gender' (override with `.groups` argument)

##     1   2   3
## F 155 170 155
## M 175 175 190
## # A tibble: 2 x 4
## # Groups:   gender [2]
##   gender   `1`   `2`   `3`
##   <chr>  <dbl> <dbl> <dbl>
## 1 F        155   170   155
## 2 M        175   175   190

aggregate(h ~ gender + num, sum, data=dat)
dat %>% group_by(gender, num) %>% summarise(h=sum(h)) %>% arrange(num)

## `summarise()` regrouping output by 'gender' (override with `.groups` argument)

##   gender num   h
## 1      F   1 310
## 2      M   1 350
## 3      F   2 170
## 4      M   2 350
## 5      F   3 310
## 6      M   3 190
## # A tibble: 6 x 3
## # Groups:   gender [2]
##   gender   num     h
##   <chr>  <dbl> <dbl>
## 1 F          1   310
## 2 M          1   350
## 3 F          2   170
## 4 M          2   350
## 5 F          3   310
## 6 M          3   190

aggregate(h + w ~ gender + num, sum, data=dat)
dat %>% group_by(gender, num) %>% summarise(h=sum(h+w)) %>% arrange(num)

## `summarise()` regrouping output by 'gender' (override with `.groups` argument)

##   gender num h + w
## 1      F   1   410
## 2      M   1   510
## 3      F   2   230
## 4      M   2   480
## 5      F   3   420
## 6      M   3   290
## # A tibble: 6 x 3
## # Groups:   gender [2]
##   gender   num     h
##   <chr>  <dbl> <dbl>
## 1 F          1   410
## 2 M          1   510
## 3 F          2   230
## 4 M          2   480
## 5 F          3   420
## 6 M          3   290

위의 경우 h=sum(h+w)인지 h=sum(h)+sum(w)인지 헷갈릴 수 있다. h=max(h+w)와 h=max(h)+max(w)는 다르기 때문에 다음에서 확인하자.

aggregate(h + w ~ gender + num, max, data=dat) 
dat %>% group_by(gender, num) %>% summarise(`h + w`=max(h+w)) %>% arrange(num)

## `summarise()` regrouping output by 'gender' (override with `.groups` argument)

##   gender num h + w
## 1      F   1   210
## 2      M   1   260
## 3      F   2   230
## 4      M   2   250
## 5      F   3   220
## 6      M   3   290
## # A tibble: 6 x 3
## # Groups:   gender [2]
##   gender   num `h + w`
##   <chr>  <dbl>   <dbl>
## 1 F          1     210
## 2 M          1     260
## 3 F          2     230
## 4 M          2     250
## 5 F          3     220
## 6 M          3     290

aggregate(cbind(h,w)~gender+num, sum, data=dat)
dat %>% group_by(gender, num) %>% summarise(h=sum(h), w=sum(w)) %>% arrange(num)

## `summarise()` regrouping output by 'gender' (override with `.groups` argument)

##   gender num   h   w
## 1      F   1 310 100
## 2      M   1 350 160
## 3      F   2 170  60
## 4      M   2 350 130
## 5      F   3 310 110
## 6      M   3 190 100
## # A tibble: 6 x 4
## # Groups:   gender [2]
##   gender   num     h     w
##   <chr>  <dbl> <dbl> <dbl>
## 1 F          1   310   100
## 2 M          1   350   160
## 3 F          2   170    60
## 4 M          2   350   130
## 5 F          3   310   110
## 6 M          3   190   100

aggregate(. ~ gender + num, sum, data=dat)
dat %>% group_by(gender, num) %>% summarise_all(sum) %>% arrange(num)

##   gender num   h   w      BMI
## 1      F   1 310 100 41.75347
## 2      M   1 350 160 52.37302
## 3      F   2 170  60 20.76125
## 4      M   2 350 130 42.36618
## 5      F   3 310 110 45.65972
## 6      M   3 190 100 27.70083
## # A tibble: 6 x 5
## # Groups:   gender [2]
##   gender   num     h     w   BMI
##   <chr>  <dbl> <dbl> <dbl> <dbl>
## 1 F          1   310   100  41.8
## 2 M          1   350   160  52.4
## 3 F          2   170    60  20.8
## 4 M          2   350   130  42.4
## 5 F          3   310   110  45.7
## 6 M          3   190   100  27.7

aggregate(dat, list(dat$gender, dat$num), length)
dat %>% group_by(gender, num) %>% summarise_all(length) %>% arrange(num)

##   Group.1 Group.2 gender num h w BMI
## 1       F       1      2   2 2 2   2
## 2       M       1      2   2 2 2   2
## 3       F       2      1   1 1 1   1
## 4       M       2      2   2 2 2   2
## 5       F       3      2   2 2 2   2
## 6       M       3      1   1 1 1   1
## # A tibble: 6 x 5
## # Groups:   gender [2]
##   gender   num     h     w   BMI
##   <chr>  <dbl> <int> <int> <int>
## 1 F          1     2     2     2
## 2 M          1     2     2     2
## 3 F          2     1     1     1
## 4 M          2     2     2     2
## 5 F          3     2     2     2
## 6 M          3     1     1     1

by(dat, list(dat$gender, dat$num), summary)
# dplyr로 구현하기 힘듦
require(data.table)
datDT <- data.table(dat)
datDT[,{print(paste(gender, num));print(summary(.SD))}, by=c('gender', 'num')]

## : F
## : 1
##     gender               num          h               w           BMI       
##  Length:2           Min.   :1   Min.   :150.0   Min.   :50   Min.   :19.53  
##  Class :character   1st Qu.:1   1st Qu.:152.5   1st Qu.:50   1st Qu.:20.20  
##  Mode  :character   Median :1   Median :155.0   Median :50   Median :20.88  
##                     Mean   :1   Mean   :155.0   Mean   :50   Mean   :20.88  
##                     3rd Qu.:1   3rd Qu.:157.5   3rd Qu.:50   3rd Qu.:21.55  
##                     Max.   :1   Max.   :160.0   Max.   :50   Max.   :22.22  
## --------------------------------------------------------------------------------- 
## : M
## : 1
##     gender               num          h               w           BMI       
##  Length:2           Min.   :1   Min.   :170.0   Min.   :80   Min.   :24.69  
##  Class :character   1st Qu.:1   1st Qu.:172.5   1st Qu.:80   1st Qu.:25.44  
##  Mode  :character   Median :1   Median :175.0   Median :80   Median :26.19  
##                     Mean   :1   Mean   :175.0   Mean   :80   Mean   :26.19  
##                     3rd Qu.:1   3rd Qu.:177.5   3rd Qu.:80   3rd Qu.:26.93  
##                     Max.   :1   Max.   :180.0   Max.   :80   Max.   :27.68  
## --------------------------------------------------------------------------------- 
## : F
## : 2
##     gender               num          h             w           BMI       
##  Length:1           Min.   :2   Min.   :170   Min.   :60   Min.   :20.76  
##  Class :character   1st Qu.:2   1st Qu.:170   1st Qu.:60   1st Qu.:20.76  
##  Mode  :character   Median :2   Median :170   Median :60   Median :20.76  
##                     Mean   :2   Mean   :170   Mean   :60   Mean   :20.76  
##                     3rd Qu.:2   3rd Qu.:170   3rd Qu.:60   3rd Qu.:20.76  
##                     Max.   :2   Max.   :170   Max.   :60   Max.   :20.76  
## --------------------------------------------------------------------------------- 
## : M
## : 2
##     gender               num          h               w             BMI       
##  Length:2           Min.   :2   Min.   :170.0   Min.   :60.0   Min.   :20.76  
##  Class :character   1st Qu.:2   1st Qu.:172.5   1st Qu.:62.5   1st Qu.:20.97  
##  Mode  :character   Median :2   Median :175.0   Median :65.0   Median :21.18  
##                     Mean   :2   Mean   :175.0   Mean   :65.0   Mean   :21.18  
##                     3rd Qu.:2   3rd Qu.:177.5   3rd Qu.:67.5   3rd Qu.:21.39  
##                     Max.   :2   Max.   :180.0   Max.   :70.0   Max.   :21.60  
## --------------------------------------------------------------------------------- 
## : F
## : 3
##     gender               num          h               w             BMI       
##  Length:2           Min.   :3   Min.   :150.0   Min.   :50.0   Min.   :22.22  
##  Class :character   1st Qu.:3   1st Qu.:152.5   1st Qu.:52.5   1st Qu.:22.53  
##  Mode  :character   Median :3   Median :155.0   Median :55.0   Median :22.83  
##                     Mean   :3   Mean   :155.0   Mean   :55.0   Mean   :22.83  
##                     3rd Qu.:3   3rd Qu.:157.5   3rd Qu.:57.5   3rd Qu.:23.13  
##                     Max.   :3   Max.   :160.0   Max.   :60.0   Max.   :23.44  
## --------------------------------------------------------------------------------- 
## : M
## : 3
##     gender               num          h             w            BMI      
##  Length:1           Min.   :3   Min.   :190   Min.   :100   Min.   :27.7  
##  Class :character   1st Qu.:3   1st Qu.:190   1st Qu.:100   1st Qu.:27.7  
##  Mode  :character   Median :3   Median :190   Median :100   Median :27.7  
##                     Mean   :3   Mean   :190   Mean   :100   Mean   :27.7  
##                     3rd Qu.:3   3rd Qu.:190   3rd Qu.:100   3rd Qu.:27.7  
##                     Max.   :3   Max.   :190   Max.   :100   Max.   :27.7  
## [1] "M 1"
##        h               w           BMI       
##  Min.   :170.0   Min.   :80   Min.   :24.69  
##  1st Qu.:172.5   1st Qu.:80   1st Qu.:25.44  
##  Median :175.0   Median :80   Median :26.19  
##  Mean   :175.0   Mean   :80   Mean   :26.19  
##  3rd Qu.:177.5   3rd Qu.:80   3rd Qu.:26.93  
##  Max.   :180.0   Max.   :80   Max.   :27.68  
## [1] "M 2"
##        h               w             BMI       
##  Min.   :170.0   Min.   :60.0   Min.   :20.76  
##  1st Qu.:172.5   1st Qu.:62.5   1st Qu.:20.97  
##  Median :175.0   Median :65.0   Median :21.18  
##  Mean   :175.0   Mean   :65.0   Mean   :21.18  
##  3rd Qu.:177.5   3rd Qu.:67.5   3rd Qu.:21.39  
##  Max.   :180.0   Max.   :70.0   Max.   :21.60  
## [1] "M 3"
##        h             w            BMI      
##  Min.   :190   Min.   :100   Min.   :27.7  
##  1st Qu.:190   1st Qu.:100   1st Qu.:27.7  
##  Median :190   Median :100   Median :27.7  
##  Mean   :190   Mean   :100   Mean   :27.7  
##  3rd Qu.:190   3rd Qu.:100   3rd Qu.:27.7  
##  Max.   :190   Max.   :100   Max.   :27.7  
## [1] "F 3"
##        h               w             BMI       
##  Min.   :150.0   Min.   :50.0   Min.   :22.22  
##  1st Qu.:152.5   1st Qu.:52.5   1st Qu.:22.53  
##  Median :155.0   Median :55.0   Median :22.83  
##  Mean   :155.0   Mean   :55.0   Mean   :22.83  
##  3rd Qu.:157.5   3rd Qu.:57.5   3rd Qu.:23.13  
##  Max.   :160.0   Max.   :60.0   Max.   :23.44  
## [1] "F 1"
##        h               w           BMI       
##  Min.   :150.0   Min.   :50   Min.   :19.53  
##  1st Qu.:152.5   1st Qu.:50   1st Qu.:20.20  
##  Median :155.0   Median :50   Median :20.88  
##  Mean   :155.0   Mean   :50   Mean   :20.88  
##  3rd Qu.:157.5   3rd Qu.:50   3rd Qu.:21.55  
##  Max.   :160.0   Max.   :50   Max.   :22.22  
## [1] "F 2"
##        h             w           BMI       
##  Min.   :170   Min.   :60   Min.   :20.76  
##  1st Qu.:170   1st Qu.:60   1st Qu.:20.76  
##  Median :170   Median :60   Median :20.76  
##  Mean   :170   Mean   :60   Mean   :20.76  
##  3rd Qu.:170   3rd Qu.:60   3rd Qu.:20.76  
##  Max.   :170   Max.   :60   Max.   :20.76  
## Empty data.table (0 rows and 2 cols): gender,num