后续:泛化一个data.frame子集功能2

问题描述

我正在跟进这个很棒的答案.在那个答案中，foo2 函数帮助用户识别在哪个唯一的 study、另外两个选定的列(group&outcome) 是恒定或变化.

I'm following up on this great answer. In that answer, the foo2 function helped the user identify in which unique study, any of the two other selected columns (group&outcome) are constant or vary.

现在，假设我们要另外确定是否存在唯一的study，其中所选变量之一(结果) 对于任何其他选定变量 (group) 的某些行来说是常量.

Now, imagine we want to additionally identify if there are unique study in which one of the selected variables (outcome) is constant for some rows of any other selected variables (group).

例如，在study==14中，outcome对于group的某些行是恒定的.但是在study==8中，outcome对于group的所有行都完全不同:

For example, in study==14, outcome is constant for some rows of group. But in study==8, outcome is completely varying for all rows of group:

   study group outcome
17    14     1       6
18    14     2       6
19    14     3       7
20    14     4       7

   study group outcome
9      8     1       2
10     8     2       3
11     8     3       4

有没有办法将foo2扩展到附加识别像study==14这样的研究?

Is there a way to extend foo2 to additionally identify studies like study==14?

dat = read.csv("https://raw.githubusercontent.com/rnorouzian/s/main/cf.csv")

study8 = subset(dat, study==8)[1:3]; study14 = subset(dat, study==14)[1:3] 

推荐答案

我们可以用 case_when

foo2 <- function(dat, study_col, ...) {
   
    dot_cols <- ensyms(...)
    str_cols <- purrr::map_chr(dot_cols, rlang::as_string)

    dat %>%
           dplyr::select({{study_col}}, !!! dot_cols) %>%
            dplyr::group_by({{study_col}}) %>%
            dplyr::mutate(grp = across(all_of(str_cols), ~ {
                      tmp <- n_distinct(.)
                      case_when(tmp  == 1 ~ 1, tmp == n() ~ 2, tmp >1 & tmp < n() ~ 3,  TRUE ~ 4)
                      }) %>%
                   purrr::reduce(stringr::str_c, collapse="")) %>%
                dplyr::ungroup(.) %>%
                 dplyr::group_split(grp, .keep = FALSE)
           
     }

foo2(dat, study, group, outcome)

这篇关于后续:泛化一个data.frame子集功能2的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！