我怎样才能编写一个 tidyverse 友好的函数，在管道中更早地尊重 group_by() ？

Question

我已经开始着手编写函数以加快 table 生成速度，但希望该函数尊重用户在管道中所做的早期分组选择。

示例数据：

df<-data.frame(ID=c("A","B","C","A","C","D","A","C","E","B","C","A"),
           Year=c(1,1,1,2,2,2,3,3,3,4,4,4),
           Credits=c(1,3,4,5,6,7,2,1,1,6,1,2),
           Major=c("GS","GS","LA","GS","GS","LA","GS","LA","LA","GS","LA","LA"),
           Status=c("green","blue","green","blue","green","blue","green","blue","green","blue","green","blue"),
           Group=c("Art","Music","Science","Art","Music","Science","Art","Music","Science","Art","Music","Science"))

以下是我正在处理的函数，它 requires/accepts 一个定义群组的变量、一个信用变量和一个期限变量。

table_headsfte_cohorts<-function(.data,cohortvar,credits,term){


  cohortvar<-rlang::ensym(cohortvar)
  credits<-rlang::ensym(credits)
  term<-rlang::ensym(term)


  .data%>%
    group_by(!!term,Pidm)%>%
    group_by(!!term,!!cohortvar,group_cols())%>%
    mutate(on3=1)%>%
    mutate(`Headcount`=sum(on3),
          `FTE`=round(sum(na.omit(!!credits))/15,1))%>%
    mutate(Variable=paste0(cohortvar))%>%
    mutate(Category=!!cohortvar)%>%
    select(-!!cohortvar)%>%
    select(Variable,Category,Headcount,FTE,group_cols())
}

对于可能有兴趣使用附加分组变量的用户，我希望最终结果函数允许使用如下：

df2<-df%>%
 group_by(Status,Group)%>%
 table_headsfte_cohorts(Major,Credits,Year)

期望的最终结果将是 table，除了 cohortvar 和 [=20] 之外，尊重并保留上述 group_by 语句中两个分组变量的水平=] 列来自 table_headsfte_cohorts() 个参数。

我需要生成相同的 table，但对于范围广泛的分组变量和不同数量的分组变量，灵活性将非常有帮助。

编辑：

以下似乎接近了，至少允许多个分组变量。这不是我所希望的，因为我更喜欢从管道中读取额外的分组参数：

 table_headsfte_cohorts<-function(.data,cohortvar,credits,term,...){

  grps<-enquos(...)

  cohortvar<-rlang::ensym(cohortvar)
  credits<-rlang::ensym(credits)
  term<-rlang::ensym(term)


  .data%>%
      group_by(!!term,!!cohortvar,!!! grps)%>%
     mutate(on3=1)%>%
     mutate(`Headcount`=sum(on3),
          `FTE`=round(sum(na.omit(!!credits))/15,1))%>%
     mutate(Variable=paste0(cohortvar))%>%
     mutate(Category=!!cohortvar)%>%
     select(-!!cohortvar)%>%
     select(Variable,Category,Headcount,FTE,!!!grps)

}

利用上面的，我可以成功运行:

fdfout<-fdf%>%
table_headsfte_cohorts(Major, Credits, Year), getting:

我还可以将其他变量传递给函数作为附加分组变量：

fdfout_alt<-fdf%>%
  table_headsfte_cohorts(Major,Credits,Year,Status,Group)

产生了想要的结果：

不幸的是，当我使用

fdf_no<-fdf%>%
  group_by(Status, Group)%>%
  table_headsfte_cohorts(Major, Credits, Year)

我得到：

此输出可能会使使用我的函数的人感到困惑，因为他们的 group_by() 行似乎什么都不做。

Answer 1

我添加了一些行，将点内的现有分组变量和新分组变量合并到一个字符向量中。我们可以用 group_vars 得到现有的分组变量。要将新旧合并在一起，我们必须获取引用分组变量的表达式 get_expr 并将它们转换为字符串。我们可以使用 !!! syms 来评估和 all_of 到 select 分组变量。

这是你的想法吗？

table_headsfte_cohorts <- function(.data, cohortvar, credits, term, ...){
  
  new_grps <- enquos(...)
  new_grps <- purrr::map_chr(new_grps, ~ as.character(rlang::get_expr(.x)))
  ex_grps  <- group_vars(.data)
  grp_vars <- c(ex_grps, new_grps)

  cohortvar<-rlang::ensym(cohortvar)
  credits<-rlang::ensym(credits)
  term<-rlang::ensym(term)
  
  
  .data%>%
    group_by(!! term,
             !! cohortvar,
             !!! syms(grp_vars))%>%
    mutate(on3 = 1) %>%
    mutate(`Headcount`= sum(on3),
           `FTE`= round(sum(na.omit(!!credits))/15,1))%>%
    mutate(Variable=paste0(cohortvar))%>%
    mutate(Category=!!cohortvar)%>%
    select(-!!cohortvar)%>%
    select(Variable,Category,Headcount,FTE, all_of(grp_vars))
  
}

df %>%
  group_by(Status, Group) %>%
  table_headsfte_cohorts(Major, Credits, Year)

#> Adding missing grouping variables: `Major`
#> Adding missing grouping variables: `Year`, `Major`
#> # A tibble: 12 x 8
#> # Groups:   Year, Major, Status, Group [12]
#>     Year Major Variable Category Headcount   FTE Status Group  
#>    <dbl> <chr> <chr>    <chr>        <dbl> <dbl> <chr>  <chr>  
#>  1     1 GS    Major    GS               1   0.1 green  Art    
#>  2     1 GS    Major    GS               1   0.2 blue   Music  
#>  3     1 LA    Major    LA               1   0.3 green  Science
#>  4     2 GS    Major    GS               1   0.3 blue   Art    
#>  5     2 GS    Major    GS               1   0.4 green  Music  
#>  6     2 LA    Major    LA               1   0.5 blue   Science
#>  7     3 GS    Major    GS               1   0.1 green  Art    
#>  8     3 LA    Major    LA               1   0.1 blue   Music  
#>  9     3 LA    Major    LA               1   0.1 green  Science
#> 10     4 GS    Major    GS               1   0.4 blue   Art    
#> 11     4 LA    Major    LA               1   0.1 green  Music  
#> 12     4 LA    Major    LA               1   0.1 blue   Science

我怎样才能编写一个 tidyverse 友好的函数，在管道中更早地尊重 group_by() ？

How can I write a tidyverse-friendly function that respects group_by() earlier in the pipe?

r

dplyr

tidyr