问题描述

数据框中包含city和dim两个分类变量，以及area和weight两个测量值，想要先基于city进行分类，在此基础上计算各个小组中dim为特定类别的条目的area或weight统计值。我用dplyr的group_by和summarise能完成计算，想知道如何把这个计算过程封装成一个函数呢？具体说明如下。

代码和结果

# 载入需要的包
library(dplyr)

# 构建数据
mydata <- data.frame(
  "city" = rep(c("Beijing", "Shanghai", "Kunming"), each = 4), 
  "dim" = rep(c(1,2,3), 4), 
  "area" = seq(1:12),
  "weight" = rep(2, 12)
)

这是我的计算过程和应得的结果：

# 按city分组，然后分别统计各个分组中dim = 1的area占组内area总值的比例
# 然后再分别统计各组中dim = 1的weight占组内weight总值的比例
mydata %>% group_by(city) %>% 
  summarise(perc = sum(ifelse(dim == 1, area, 0)/sum(area)))
mydata %>% group_by(city) %>% 
  summarise(perc = sum(ifelse(dim == 1, weight, 0)/sum(weight)))

# `summarise()` ungrouping output (override with `.groups` argument)
#  A tibble: 3 x 2
#   city      perc
#   <chr>    <dbl>
# 1 Beijing  0.5  
# 2 Kunming  0.238
# 3 Shanghai 0.269

# `summarise()` ungrouping output (override with `.groups` argument)
#  A tibble: 3 x 2
#   city      perc
#   <chr>    <dbl>
# 1 Beijing   0.5 
# 2 Kunming   0.25
# 3 Shanghai  0.25

然后尝试构建可以实现上一步分组统计的函数：其中x为输入数据框mydata，y为要统计的变量，即area或者weight：

myfun <- function(x,y) {
  x %>% group_by(city) %>% 
    summarise(perc = sum(ifelse(dim == 1, y, 0)/sum(y)))
}
myfun(mydata, mydata$area)
# 然而结果错误：

# `summarise()` ungrouping output (override with `.groups` argument)
#  A tibble: 3 x 2
#   city       perc
#   <chr>     <dbl>
# 1 Beijing  0.0641
# 2 Kunming  0.0256
# 3 Shanghai 0.0385

# 于是尝试这么搞：
myfun(mydata, "area")
# 然而提示运行错误：

#  Error: Problem with `summarise()` input `perc`.
# x invalid 'type' (character) of argument
# i Input `perc` is `sum(ifelse(dim == 1, y, 0)/sum(y))`.
# i The error occurred in group 1: city = "Beijing".
# Run `rlang::last_error()` to see where the error occurred.