如何对data.table执行melt时仅对measure.vars首元素应用na.rm？

The core issue here is that data.table::melt()'s na.rm = TRUE removes rows where any of the measure variables have NA values—including your level_date column, which you want to keep even if it's NA (as long as level_id is valid).

The cleanest, most efficient solution is to first perform the melt as you originally did, then filter out only the rows where level_id is NA using data.table's native row indexing. This avoids hacky workarounds like fake date values and keeps your code readable.

Step-by-Step Implementation #

1. First, recreate your original data.table (for reproducibility):

library(data.table)
library(lubridate)

dt.master <- data.table(user = seq(1,5),
                        visit_id = c(2,4,NA,4,8),
                        visit_date = c(dmy("10/02/2018"), dmy("11/04/2018"), NA, dmy("02/03/2018"), NA),
                        offer_id = c(1,3,NA,NA,NA),
                        offer_date = c(dmy("15/02/2018"), dmy("18/04/2018"), NA, NA, NA))

2. Perform the melt, then chain a filter to remove rows where level_id is NA:

dt.melted <- melt(dt.master, id.vars = "user",
                  measure.vars = list(c("visit_id", "offer_id"), c("visit_date", "offer_date")),
                  variable.name = "level", value.name = c("level_id", "level_date"))[!is.na(level_id)]

Result Verification #

Running this code gives you exactly the output you want:

user level level_id level_date
1:    1     1        2 2018-02-10
2:    2     1        4 2018-04-11
3:    4     1        4 2018-03-02
4:    5     1        8       <NA>
5:    1     2        1 2018-02-15
6:    2     2        3 2018-04-18

Why This Works Better Than na.rm = TRUE #

• The default na.rm = TRUE in melt() would drop the row for user 5 (level 1) because level_date is NA—this solution preserves it, since we only care about level_id being non-NA.
• It's idiomatic data.table code: chaining operations keeps your logic concise and efficient, avoiding unnecessary intermediate objects.
• No messy workarounds (like filling fake dates) are needed, so your code stays maintainable.

内容的提问来源于stack exchange，提问作者illanxr