有没有办法根据条件将相关变量折叠成一个？

Question

比方说，我有多个变量衡量物质滥用即A1是alcohal使用， A2是bhang和A3是可卡因。我想生成变量afin如果三者中的任何一个是肯定的，表明参与药物滥用的。

有没有办法缩短代码，所以我不指定使用多个ifelse语句，如下所示？试图找到最好的方法来做到这一点，因为我有 10 多个变量可以合并为一个，而写作ifelse可能并不理想。

# Anymatch
library(tidyverse)

set.seed(2021)

mydata <- tibble(
  a1 = factor(round(runif(20, 1, 3)),
              labels = c("Yes", "No", "N/A")),
  a2 = factor(round(runif(20, 1, 3)),
              labels = c("Yes", "No", "N/A")),
  a3 = factor(round(runif(20, 1, 3)),
              labels = c("Yes", "No", "N/A")),
  b1 = round(rnorm(20, 10, 2)))
mydata

mydata <- mydata %>%
  mutate(afin = ifelse(a1 == "Yes"|a2=="Yes"|a3=="Yes", "Yes", "No"))

Answer 1

我们也可以在没有 a 的情况下做到这一点ifelse。只需将逻辑列转换为数字索引并传递 avector来替换值

library(dplyr)
mydata %>%
     mutate(afin = c("no", "yes")[1 + (rowSums(select(cur_data(), 
        starts_with('a')) == 'Yes') > 0)])

-输出

# A tibble: 20 x 5
   a1    a2    a3       b1 afin 
   <fct> <fct> <fct> <dbl> <chr>
 1 No    Yes   Yes       6 yes  
 2 N/A   N/A   N/A       7 no   
 3 No    No    No       12 no   
 4 No    No    N/A       7 no   
 5 No    No    Yes       9 yes  
 6 No    N/A   N/A       7 no   
 7 No    N/A   N/A       7 no   
 8 No    N/A   Yes       7 yes  
 9 N/A   N/A   Yes      10 yes  
10 N/A   N/A   N/A      11 no   
11 Yes   Yes   No       10 yes  
12 N/A   N/A   No       14 no   
13 No    N/A   Yes       9 yes  
14 No    N/A   No       14 no   
15 N/A   No    No       10 no   
16 No    Yes   Yes       8 yes  
17 No    N/A   No       13 no   
18 N/A   Yes   No        9 yes  
19 N/A   N/A   N/A      11 no   
20 No    No    N/A      11 no   

或使用 c_across

mydata %>% 
   rowwise %>%
   mutate(afin = c("no", "yes")[1+ 
          any(c_across(starts_with('a')) == "Yes")]) %>% 
   ungroup

Answer 2

您可以使用rowSums-

library(dplyr)

mydata %>% 
   mutate(afin = ifelse(rowSums(select(., a1:a3) == 'Yes') > 0, 'yes', 'no'))

#   a1    a2    a3       b1 afin 
#   <fct> <fct> <fct> <dbl> <chr>
# 1 No    Yes   Yes       6 yes  
# 2 N/A   N/A   N/A       7 no   
# 3 No    No    No       12 no   
# 4 No    No    N/A       7 no   
# 5 No    No    Yes       9 yes  
# 6 No    N/A   N/A       7 no   
# 7 No    N/A   N/A       7 no   
# 8 No    N/A   Yes       7 yes  
# 9 N/A   N/A   Yes      10 yes  
#10 N/A   N/A   N/A      11 no   
#11 Yes   Yes   No       10 yes  
#12 N/A   N/A   No       14 no   
#13 No    N/A   Yes       9 yes  
#14 No    N/A   No       14 no   
#15 N/A   No    No       10 no   
#16 No    Yes   Yes       8 yes  
#17 No    N/A   No       13 no   
#18 N/A   Yes   No        9 yes  
#19 N/A   N/A   N/A      11 no   
#20 No    No    N/A      11 no