使用ifelse()生成新变量时丢失日期格式问题排查

I totally get this frustration—base R's ifelse() has a quirk where it strips away the Date class from your results. That's because Date objects are actually stored as integers under the hood (counting days since 1970-01-01), and ifelse() coerces the output to a raw integer vector, losing the date formatting metadata.

Here are three solid solutions to fix this:

Solution 1: Use dplyr::if_else() (Recommended) #

The if_else() function from the dplyr package is built to preserve data types when both the true and false outputs are the same class. Perfect for dates!

First, load dplyr if you haven't already:

library(dplyr)

Then create your new date variable:

x <- x %>%
  mutate(matching_date = if_else(DOB == SU_BIRTH_DATE, DOB, SU_BIRTH_DATE))

Check the structure to confirm the date class is retained:

glimpse(x)

You'll see matching_date still shows up as <date>.

Solution 2: Manually Convert Back to Date Class (Base R Only) #

If you want to stick with base R's ifelse(), just wrap the result in as.Date() to restore the class. Don't forget the origin parameter—it tells R what the integer values correspond to (the standard date origin is 1970-01-01):

x$matching_date <- as.Date(
  ifelse(x$DOB == x$SU_BIRTH_DATE, x$DOB, x$SU_BIRTH_DATE),
  origin = "1970-01-01"
)

Solution 3: Use case_when() for Complex Conditions #

If you have multiple conditional checks, dplyr::case_when() is a clean, readable option that also preserves the Date class:

x <- x %>%
  mutate(matching_date = case_when(
    DOB == SU_BIRTH_DATE ~ DOB,
    DOB > SU_BIRTH_DATE ~ SU_BIRTH_DATE,
    TRUE ~ DOB  # Catch-all for remaining cases
  ))

内容的提问来源于stack exchange，提问作者Doug Fir