我们需要 na .rm = FALSE 。默认情况下，它是 na.rm = TRUE ，它会在开始时删除 NA 值，以便输出具有比原始数据集中的元素数量少。 dplyr :: mutate 仅在输出元素的长度与原始数据集的 length 相同时有效。

We need the na.rm=FALSE. By default, it is na.rm=TRUE which removes the NA values at the beginning so that the output will have less number of elements than in the original dataset. The dplyr::mutate will only work when the output elements have the same length as the original dataset.

library(zoo)
library(dplyr)
sample %>%
     group_by(ID) %>% 
     mutate(Date = na.locf(as.Date(Date, origin = '1970-01-01'), na.rm=FALSE))
#     ID       Date
#  (int)     (date)
#1     1       <NA>
#2     1       <NA>
#3     1 2015-01-01
#4     1 2015-01-01
#5     2       <NA>
#6     2 2015-01-01
#7     2 2015-01-01
#8     2 2015-01-01