如何计算嵌套组的平均值并计算 R 中的观察次数

Question

all (R users), THANK YOU VERY MUCH in advance. I've a data set that contains students' scores from multiple states. Each state has different schools (10 schools in this example), each school has to be either 'public' or 'private'; and three items' test scores. I need to calculate the mean of each school for each item, and display the type of school, then save the results into excel file to export them.

The expected result of the excel file would include:

1. column of the name of the state,
2. column of the name of the schools (10 schools for each state column),
3. column of the type of the school (to indicate 'public' or 'private'),
4. number of students in each school,
5. mean of item1,
6. mean of item2, and
7. mean of item3.

library(randomNames)

# example to demonstrate the general concept): 
ID = 1:50
states = rep(c("TS", "NE", "AR", "MO", "WA"),times = c(10, 10, 10, 10, 10))
schools = randomNames::randomNames(50) ## 5 first last names separated by a space
type = rep(c("private", "public"),times = c(20,30))
item1 = rnorm(50, mean=25, sd=5)
item2 = rnorm(50, mean=30, sd=5)
item3 = rnorm(50, mean=15, sd=5)
df = data.frame(ID, states, schools, type, item1, item2, item3)

Then I need to save it into excel file to export each state separately using the following code:

# this below code works fine, I'm just adding it to explain the full concept. 

list_data <- split(df, df$states)
Map(openxlsx::write.xlsx, list_data, paste0(names(list_data), '.xlsx'))

THANK YOU VERY MUCH.

Answer 1

You can do this with dplyr and tidyr packages:

library(dplyr)
library(tidyr)

df %>% 
  dplyr::group_by(states, schools, type) %>% 
  dplyr::summarize(across(tidyr::starts_with("item"), ~ mean(.)),
                   students = n()) %>%
  dplyr::ungroup()

   states schools             type   item1 item2 item3 students
   <chr>  <chr>               <chr>  <dbl> <dbl> <dbl>    <int>
 1 AR     al-Hosein, Zubaida  public  23.4  35.1 15.4         1
 2 AR     al-Mohamed, Raadiya public  24.5  30.8 13.5         1
 3 AR     Bluford, Sage       public  29.9  32.4  9.49        1
 4 AR     Covarrubias, Julio  public  19.8  27.8 15.2         1
 5 AR     el-Gad, Naaila      public  27.0  33.5 19.5         1
 6 AR     el-Mansour, Fawzia  public  34.4  25.4 17.9         1
 7 AR     el-Sadri,  Sakeena  public  24.7  30.5 13.9         1
 8 AR     Ewers, Benjamin     public  18.3  33.6 13.5         1
 9 AR     Rivas, Joel         public  16.8  25.1 20.5         1
10 AR     Wilson, Reneisha    public  28.9  28.5 18.5         1
# ... with 40 more rows

If you have other column names that start with item then you can change the line across(tidyr::starts_with(.... to item1 = mean(item1) and so on.

The count for student assumes that each row within a school and state is a student and that the type does not change for a given school.