r - 按行识别匹配项，并使用匹配项的列名创建列

Question

我一直在尝试(未成功)使用来自多个线程的解决方案，尤其是: Get column name based on row values matching in DF in R (more than 3 column names) Populate a new column if a value is found in any column但是当我使它们适应我的特定需求时，我失败了……其他帖子处理与特定值或最大值等的匹配，但这对我不起作用。

所以我有一个变量(日期)，我需要根据其他日期变量知道该日期的来源。我有一个非常大的数据集，所以我需要保留其余部分，并避免使用对列号的引用(例如 df[2:3])，因为我的数据源将来可能会发生变化，所以最好坚持到变量名称(例如，“date1”)。最好举例说明:

df<-data.frame(id=as.character(c(1,2,3,4,5)), 
               date1=c("01/10/2022",NA,"1/10/2022","2/10/2022",NA),
               date2=c("02/10/2022",NA,"3/10/2022","4/10/2022","02/10/2022"),
               date3=c("05/10/2022","3/10/2022", "5/10/2022","6/10/2022","04/10/2022"),
               keydate=c("01/10/2022","3/10/2022", "1/10/2022","4/10/2022","05/10/2022"))

df$date1 <- as.Date(df$date1, format="%d/%m/%y")
df$date2 <- as.Date(df$date2, format="%d/%m/%y")
df$date3 <- as.Date(df$date3, format="%d/%m/%y")
df$keydate <- as.Date(df$keydate, format="%d/%m/%y")

然后我需要一个新变量“datesource”来定义“keydate”的来源，在我之前的例子中是这样的:

datesource=c("date1","date3","date1","date2","none")
cbind(df,datesource)

我的一些失败尝试包括:

df$datesource = ifelse(df$keydate %in% df[c(date1,date2,date3),], colnames(df), "none") #get all "none"
#OR
df %>% mutate(datesource = ifelse(df$keydate %in% df, colnames(df), "none")) #get all "none"
#OR
df$datesource <- apply(df, colnames(df[c(date1,date2,date3),]), 
                     function(x) ifelse(any(x = df$keydate), colnames(df), 'none')) 
#Error in apply(df, colnames(df[c(date1, date2, date3), ]), function(x) ifelse(any(x = df$keydate),  : 'X' must have named dimnames

谢谢你的帮助!

Answer 1

这是一个选项。您可以在列名向量中映射出正确的位置:

library(tidyverse)

df |>
  mutate(datesource = pmap_chr(list(date1, date2, date3, keydate), 
                               ~c("date1", "date2", "date3")[
                                 c(..1, ..2, ..3) %in% ..4
                               ]))
#>   id      date1      date2      date3    keydate datesource
#> 1  1 2020-10-01 2020-10-02 2020-10-05 2020-10-01      date1
#> 2  2       <NA>       <NA> 2020-10-03 2020-10-03      date3
#> 3  3 2020-10-01 2020-10-03 2020-10-05 2020-10-01      date1
#> 4  4 2020-10-02 2020-10-04 2020-10-06 2020-10-04      date2
#> 5  5       <NA> 2020-10-02 2020-10-04 2020-10-02      date2

编辑包括“无”的选项

library(tidyverse)

df |>
  mutate(datesource = pmap_chr(list(date1, date2, date3, keydate), 
                               ~ifelse(any(c(..1, ..2, ..3) %in% ..4),
                                 c("date1", "date2", "date3")[c(..1, ..2, ..3) %in% ..4],
                                 "none")))
#>   id      date1      date2      date3    keydate datesource
#> 1  1 2020-10-01 2020-10-02 2020-10-05 2020-10-01      date1
#> 2  2       <NA>       <NA> 2020-10-03 2020-10-03      date3
#> 3  3 2020-10-01 2020-10-03 2020-10-05 2020-10-01      date1
#> 4  4 2020-10-02 2020-10-04 2020-10-06 2020-10-04      date2
#> 5  5       <NA> 2020-10-02 2020-10-04 2020-10-05       none