基于列表 R 的动态 group_by 参数 for 循环/函数

Question

使用dplyr

这是我的数据集：

Class   Time    Honors  Grade    Total Students
Math    AM      Yes     PassFail    23
English AM      No      Letter      31
Science AM      Yes     Letter      22
Gym     AM      No      PassFail    26
Math    PM      Yes     PassFail    19
English PM      No      Letter      23
Science PM      Yes     Letter      24
Gym     PM      No      PassFail    13
Math    AM      Yes     PassFail    24
English AM      Yes     Letter      27
Science AM      No      Letter      28
Math    PM      No      Letter      21
English PM      Yes     PassFail    23
Science PM      No      PassFail    22

我想运行四个查询，四个越来越具体的答案。第一个查询将有一个 group_by 参数，后两个 group_by 参数，三个用于第三个，依此类推。

#query 1 
df %>%
  group_by(Class) %>%
  summarise(NewValue = mean(`Total Students`))

#results
    Class NewValue
    <chr>    <dbl>
1 English    26.00
2     Gym    19.50
3    Math    21.75
4 Science    24.0

第二个查询是相同的基本计算，但多了一个 group_by 参数。

#query2
df %>%
  group_by(Class, Time) %>%
  summarise(NewValue = mean(`Total Students`))

#results
    Class  Time NewValue
    <chr> <chr>    <dbl>
1 English    AM     29.0
2 English    PM     23.0
3     Gym    AM     26.0
4     Gym    PM     13.0
5    Math    AM     23.5
6    Math    PM     20.0
7 Science    AM     25.0
8 Science    PM     23.0

模式继续#query3将是

 df %>%
  group_by(Class, Time, Honors) %>%
  summarise(NewValue = mean(`Total Students`))

而#query4 将是

df %>%
  group_by(Class, Time, Honors, Grade) %>%
  summarise(NewValue = mean(`Total Students`))

问题：

有没有办法编写一个查询并使用 for 循环在 group_by 参数中合并不断增加的详细级别？

例如，下面的伪代码不起作用；我希望有一个类似的解决方案：

resultsarray <- data.frame()
Groupbysteps <- c( "Class", 
                   "Class, Time", 
                   "Class, Time, Honors", 
                   "Class, Time, Honors, Grade")

for (i in Groupbysteps) {
      resultsarray <- df%>%
                       group_by( Groupbysteps) %>%
                       summarise(NewValue = mean(`Total Students`))

 all <- rbind.fill(all, resultsarray)
}

Answer 1

这可以工作。

Groupbysteps <- c( "Class", "Time", "Honors", "Grade")

for (i in 1 : length(Groupbysteps)) {
      resultsarray <- df%>%
                       group_by(.dots = Groupbysteps[1 : i]) %>%
                       summarise(NewValue = mean(`Total Students`))

 all <- rbind.fill(all, resultsarray)
}

以下作品：

示例数据集

df <- iris[1:20, ]
colnames(df) <- c( "Class", "Time", "Honors", "Grade", "Total Students")
df[, 1] <- as.factor(sample(c("a", "b"), rep=T)) 
df[, 2] <- as.factor(sample(c("a", "b"), rep=T)) 
df[, 3] <- as.factor(sample(c("a", "b"), rep=T)) 
df[, 4] <- as.factor(sample(c("a", "b"), rep=T)) 
df[, 5] <- rnorm(20)

代码：

Groupbysteps <- c( "Class", "Time", "Honors", "Grade")
resultsarray <- data.frame()

for (i in 1 : length(Groupbysteps)) {

  resultsarray <- df %>%
    group_by(.dots = Groupbysteps[1 : i]) %>%
    summarise(NewValue = mean(`Total Students`))

  all <- rbind.fill(all, resultsarray)
}

Answer 2

如图所示在 rlang 中尝试syms：

library(dplyr)
library(rlang)

L <- lapply(1:4, function(i) df %>% 
                               group_by(!!!syms(names(df)[1:i])) %>% 
                               summarize(newValue = mean(Total_Students))
)

给出一个列表 L 的 4 个数据框，其列名是：

> lapply(L, names)
[[1]]
[1] "Class"    "newValue"

[[2]]
[1] "Class"    "Time"     "newValue"

[[3]]
[1] "Class"    "Time"     "Honors"   "newValue"

[[4]]
[1] "Class"    "Time"     "Honors"   "Grade"    "newValue"