使用 case_when 跨列创建新列

Question

我有一个大型数据集，其中包含许多状态列。我想创建一个包含参与者当前状态的新专栏。我正在尝试在 dplyr 中使用 case_when，但我不确定如何跨列。数据集的列太多，我无法输入每一列。这是数据示例：

library(dplyr)
problem <- tibble(name = c("sally", "jane", "austin", "mike"),
                  status1 = c("registered", "completed", "registered", "no action"),
                  status2 = c("completed", "completed", "registered", "no action"),
                  status3 = c("completed", "completed", "withdrawn", "no action"),
                  status4 = c("withdrawn", "completed", "no action", "registered"))

对于代码，我想要一个新列来说明参与者的最终状态；但是，如果他们的状态 ever 已完成，那么我希望它说完成，无论他们的最终状态是什么。对于此数据，答案如下所示：


answer <- tibble(name = c("sally", "jane", "austin", "mike"),
                 status1 = c("registered", "completed", "registered", "no action"),
                 status2 = c("completed", "completed", "registered", "no action"),
                 status3 = c("completed", "completed", "withdrawn", "no action"),
                 status4 = c("withdrawn", "completed", "no action", "registered"),
                 finalstatus = c("completed", "completed", "no action", "registered"))

此外，如果您能对您的代码进行任何解释，我将不胜感激！如果您的解决方案也可以使用 contains("status")，那将特别有用，因为在我的真实数据集中，状态列非常混乱（即 summary_status_5292019、sum_status_07012018 等） .

谢谢！

Answer 1

选项pmap

library(tidyverse)
problem %>%
     mutate(finalstatus =  pmap_chr(select(., starts_with('status')), ~ 
       case_when(any(c(...) == "completed")~ "completed",
             any(c(...) == "withdrawn") ~ "no action", 
     TRUE ~ "registered")))

Answer 2

下面是执行这种 "row matching" 操作的函数。与 case_when 类似，您可以按特定顺序放置 checks 向量，以便在找到一个元素的匹配项时，例如'completed' 在数据中，不考虑后面元素的匹配。

row_match <- function(data, checks, labels){
  matches <- match(unlist(data), checks)
  dim(matches) <- dim(data)
  labels[apply(matches, 1, min, na.rm = T)]
}

df %>% 
  mutate(final.stat = row_match(
                        data = select(df, starts_with('status')),
                        checks = c('completed', 'withdrawn', 'registered'),
                        labels = c('completed', 'no action', 'registered')))

# # A tibble: 4 x 6
#   name   status1    status2    status3   status4    final.stat
#   <chr>  <chr>      <chr>      <chr>     <chr>      <chr>     
# 1 sally  registered completed  completed withdrawn  completed 
# 2 jane   completed  completed  completed completed  completed 
# 3 austin registered registered withdrawn no action  no action 
# 4 mike   no action  no action  no action registered registered

使用 case_when 跨列创建新列

Use case_when across columns to make a new column

string

r

case-when

conditional-statements

dplyr