在自定义dplyr函数中更改结果变量的名称

时间:2017-07-13 09:43:40

标签: r function dplyr lazy-evaluation summary

背景

为了加快跨多个表生成分组摘要;正如我在dplyr工作流程中完成的大部分工作一样,我已经起草了一个生成所需指标的简单函数

# Function to generate summary table
generate_summary_tbl <- function(dataset, group_column, summary_column) {
    group_column   <- enquo(group_column)
    summary_column <- enquo(summary_column)
    dataset %>% 
        group_by(!!group_column) %>% 
        summarise(
            mean = mean(!!summary_column),
            sum  = sum(!!summary_column)
            # Other metrics that need to be generated frequently
        ) %>% 
        ungroup -> smryDta
    return(smryDta)
}

实施例

该功能可以根据需要运行:

>> mtcars %>% 
...     generate_summary_tbl(group_column = am, summary_column = mpg)
# A tibble: 2 x 3
     am     mean   sum
  <dbl>    <dbl> <dbl>
1     0 17.14737 325.8
2     1 24.39231 317.1

问题

我想,有条件地包含在结果中通过summary_column = mpg传递的列的名称。

示例结果,useColName = TRUE

使用useColName = TRUE调用时,结果应对应于:

>> mtcars %>% 
...     generate_summary_tbl(group_column = am, summary_column = mpg,
                             useColName = TRUE)
# A tibble: 2 x 3
     am     mean_am   sum_am
  <dbl>    <dbl>       <dbl>
1     0    17.14737    325.8
2     1    24.39231    317.1

不同之处在于变量名称 _am 中存在 mean_am 后缀,依此类推。

丑陋的解决方案

部分,丑陋的解决方案我使用setNames

# Function to generate summary table
generate_summary_tbl <-
    function(dataset,
             group_column,
             summary_column,
             useColName = TRUE) {
        group_column   <- enquo(group_column)
        summary_column <- enquo(summary_column)
        dataset %>%
            group_by(!!group_column) %>%
            summarise(mean = mean(!!summary_column),
                      sum  = sum(!!summary_column)) %>%
            ungroup -> smryDta

        if (useColName) {
            setNames(smryDta,
                     c(deparse(substitute(
                         group_column
                     )),
                     paste(
                         names(smryDta)[2:length(smryDta)], paste0("_", deparse(substitute(
                             group_column
                         )))
                     ))) -> smryDta
        }

        return(smryDta)
    }

实施例

返回的列名几乎匹配所需的结果。我估计我可以使用一些正则表达式并达到预期的结果。但是,我认为应该提供更有效的解决方案。

mtcars %>% 
    generate_summary_tbl(group_column = am, summary_column = mpg, useColName = TRUE)
# A tibble: 2 x 3
  `~am` `mean _~am` `sum _~am`
  <dbl>       <dbl>      <dbl>
1     0    17.14737      325.8
2     1    24.39231      317.1

如何获得所需的列名称,最好是更好地利用quolazyeval

1 个答案:

答案 0 :(得分:2)

也许使用rename

library(tidyverse)

generate_summary_tbl <- function(dataset, group_column, summary_column, useColname = FALSE) {
    group_column   <- enquo(group_column)
    summary_column <- enquo(summary_column)
    dataset %>% 
        group_by(!!group_column) %>% 
        summarise(
            mean = mean(!!summary_column),
            sum  = sum(!!summary_column)
            # Other metrics that need to be generated frequently
        ) %>% 
        ungroup -> smryDta

    if (useColname) 
      smryDta <- smryDta %>%  
      rename_at(
        vars(-one_of(quo_name(group_column))), 
        ~paste(quo_name(group_column), .x, sep="_")
      )

    return(smryDta)
}

mtcars %>% generate_summary_tbl(am, mpg)
# # A tibble: 2 x 3
#      am     mean   sum
#   <dbl>    <dbl> <dbl>
# 1     0 17.14737 325.8
# 2     1 24.39231 317.1
mtcars %>% generate_summary_tbl(am, mpg, T)
#   # A tibble: 2 x 3
#      am  am_mean am_sum
#   <dbl>    <dbl>  <dbl>
# 1     0 17.14737  325.8
# 2     1 24.39231  317.1