按组和 R 中的多列汇总

Question

I have a dataframe including some base info of students. I want to get summary statistics about Age, Sex and Group.

set.seed(500)
testdf <- data.frame(ID = paste0("Stu", c(1:10)),
                     Age = sample(18:25, 10, replace = T),
                     Sex =  sample(c("Boy", "Girl", "NA"), 10, replace = T),
                     Name = c("Pwyll","Flavian","Leehi","Zuzana","Aniya","Bogomil"
                              ,"Lameez","Prudencia","Ikuo","Grayson"),
                     GroupMath = sample(LETTERS[1:2], 10, replace = T),
                     GroupEng = sample(LETTERS[1:2], 10, replace = T),
                     GroupScie = sample(LETTERS[1:2], 10, replace = T),
                     GroupChine = sample(LETTERS[1:2], 10, replace = T))

I want it to look like this picture.(Desired output.)

And in my code I use three parts to deal with GroupMath, then GroupEng, then GroupScie, then GroupChine. Does anyone know how I can make this more efficient? I can't thank you enough.

N.math <- testdf %>% group_by(GroupMath) %>% count(GroupMath) 

Age.math <- testdf %>% group_by(GroupMath) %>% summarize(
      Mean = mean(Age),
      Max = max(Age),
      Min = min(Age),
      sd = sd(Age))


Sex.math <- testdf %>% group_by(GroupMath) %>% count(Sex)

Answer 1

Pivoting Data to Long Form

Here is at least one way to simplify your summary stats so you are not aggregating each group one-by-one. First, you can pivot your data to long format using the groups and class as your target variable, then summarise data for these groups. First, the pivot:

#### Load Tidyverse ####
library(tidyverse)

#### Pivot to Long Format ####
groups <- testdf %>%
  pivot_longer(cols = contains("Group"),
               names_to = "Class",
               values_to = "Group")
groups

Which looks like this:

# A tibble: 40 × 6
   ID      Age Sex   Name    Class      Group
   <chr> <int> <chr> <chr>   <chr>      <chr>
 1 Stu1     24 Girl  Pwyll   GroupMath  B    
 2 Stu1     24 Girl  Pwyll   GroupEng   B    
 3 Stu1     24 Girl  Pwyll   GroupScie  A    
 4 Stu1     24 Girl  Pwyll   GroupChine B    
 5 Stu2     20 Girl  Flavian GroupMath  B    
 6 Stu2     20 Girl  Flavian GroupEng   A    
 7 Stu2     20 Girl  Flavian GroupScie  A    
 8 Stu2     20 Girl  Flavian GroupChine B    
 9 Stu3     24 NA    Leehi   GroupMath  A    
10 Stu3     24 NA    Leehi   GroupEng   B 

Aggregating Data

Then you can aggregate the data using class, group, and sex:

#### Aggregate Data by Class x Group ####
sums <- groups %>% 
  group_by(Class,Group,Sex) %>% 
  summarise(
  Mean = mean(Age),
  Max = max(Age),
  Min = min(Age),
  sd = sd(Age)) %>% 
  ungroup()
sums

Shown below. Notice that some values are NA because there is only one person per gender in some cases, so there can be no standard deviation in this case:

# A tibble: 16 × 7
   Class      Group Sex    Mean   Max   Min    sd
   <chr>      <chr> <chr> <dbl> <int> <int> <dbl>
 1 GroupChine A     Boy    24      24    24 NA   
 2 GroupChine A     Girl   22.3    24    21  1.53
 3 GroupChine B     Girl   20      24    18  2.35
 4 GroupChine B     NA     24      24    24 NA   
 5 GroupEng   A     Boy    24      24    24 NA   
 6 GroupEng   A     Girl   20.5    24    19  2.38
 7 GroupEng   B     Girl   21.2    24    18  2.5 
 8 GroupEng   B     NA     24      24    24 NA   
 9 GroupMath  A     Girl   20.8    24    19  2.36
10 GroupMath  A     NA     24      24    24 NA   
11 GroupMath  B     Boy    24      24    24 NA   
12 GroupMath  B     Girl   21      24    18  2.58
13 GroupScie  A     Girl   21.7    24    20  2.08
14 GroupScie  A     NA     24      24    24 NA   
15 GroupScie  B     Boy    24      24    24 NA   
16 GroupScie  B     Girl   20.4    24    18  2.51

Then you can get gender counts like so:

#### Get Grouped Gender Counts ####
sex <- groups %>% 
  group_by(Class,Group) %>% 
  count(Sex) %>% 
  ungroup()
sex

Which looks like this:

# A tibble: 16 × 4
   Class      Group Sex       n
   <chr>      <chr> <chr> <int>
 1 GroupChine A     Boy       1
 2 GroupChine A     Girl      3
 3 GroupChine B     Girl      5
 4 GroupChine B     NA        1
 5 GroupEng   A     Boy       1
 6 GroupEng   A     Girl      4
 7 GroupEng   B     Girl      4
 8 GroupEng   B     NA        1
 9 GroupMath  A     Girl      4
10 GroupMath  A     NA        1
11 GroupMath  B     Boy       1
12 GroupMath  B     Girl      4
13 GroupScie  A     Girl      3
14 GroupScie  A     NA        1
15 GroupScie  B     Boy       1
16 GroupScie  B     Girl      5

Joining Data Frames

Finally you can join these two data frames in this way:

#### Join ####
sums %>% 
  right_join(sex)

Giving you the final product. You can see now where the NA values come from, such as Row 1 which only has 1 boy included, making SD impossible to evaluate:

Joining, by = c("Class", "Group", "Sex")
# A tibble: 16 × 8
   Class      Group Sex    Mean   Max   Min    sd     n
   <chr>      <chr> <chr> <dbl> <int> <int> <dbl> <int>
 1 GroupChine A     Boy    24      24    24 NA        1
 2 GroupChine A     Girl   22.3    24    21  1.53     3
 3 GroupChine B     Girl   20      24    18  2.35     5
 4 GroupChine B     NA     24      24    24 NA        1
 5 GroupEng   A     Boy    24      24    24 NA        1
 6 GroupEng   A     Girl   20.5    24    19  2.38     4
 7 GroupEng   B     Girl   21.2    24    18  2.5      4
 8 GroupEng   B     NA     24      24    24 NA        1
 9 GroupMath  A     Girl   20.8    24    19  2.36     4
10 GroupMath  A     NA     24      24    24 NA        1
11 GroupMath  B     Boy    24      24    24 NA        1
12 GroupMath  B     Girl   21      24    18  2.58     4
13 GroupScie  A     Girl   21.7    24    20  2.08     3
14 GroupScie  A     NA     24      24    24 NA        1
15 GroupScie  B     Boy    24      24    24 NA        1
16 GroupScie  B     Girl   20.4    24    18  2.51     5