如何根据特定的行差异添加列？

Question

我有一个data.table，格式如下：

DT <- data.table(
id=c("123", "123", "125", "125", "123", "123", "123"), 
action=c("started", "finished", "started", "finished", "started", "started", "finished"), time=c(as.POSIXct("2014-02-19 03:24:00"), as.POSIXct("2014-02-19 03:29:00"), as.POSIXct("2014-02-19 03:30:00"), as.POSIXct("2014-02-19 03:34:00"), as.POSIXct("2014-02-19 08:24:00"), as.POSIXct("2014-02-19 09:45:00"), as.POSIXct("2014-02-19 10:33:00")))

id  action      time
1   123 started     2014-02-19 03:24:00
2   123 finished    2014-02-19 03:29:00
3   125 started     2014-02-19 03:30:00
4   125 finished    2014-02-19 03:34:00
5   123 started     2014-02-19 08:24:00
6   123 started     2014-02-19 09:45:00
7   123 finished    2014-02-19 10:33:00

我想添加一个列，显示每个id行之间的时差（action：“finished” - “started”）。 该表按时间排序，但可能缺少数据（例如，可能会发生“完成” - 操作丢失，因为第5行和第6行就是这种情况。在这种情况下，应忽略第5行并且计算6和7之间的差值。最终表格应如下所示。

id  action      time                   durationInMinutes
1   123 started     2014-02-19 03:24:00    NA
2   123 finished    2014-02-19 03:29:00    5
3   125 started     2014-02-19 03:30:00    NA
4   125 finished    2014-02-19 03:34:00    4
5   123 started     2014-02-19 08:24:00    NA
6   123 started     2014-02-19 09:45:00    NA
7   123 finished    2014-02-19 10:33:00    48

有没有data.table解决方案？

Answer 1

DT[, duration := as.integer(time[action == "finished"] -
                            tail(time[action == "started"], 1))
   , by = cumsum(c(0, tail(lag(id) != id, -1)))][
     action == "started", duration := NA]
DT
#    id   action                time duration
#1: 123  started 2014-02-19 03:24:00       NA
#2: 123 finished 2014-02-19 03:29:00        5
#3: 125  started 2014-02-19 03:30:00       NA
#4: 125 finished 2014-02-19 03:34:00        4
#5: 123  started 2014-02-19 08:24:00       NA
#6: 123  started 2014-02-19 09:45:00       NA
#7: 123 finished 2014-02-19 10:33:00       48