子集data.table的每个组的前三年

Question

我有一台data.table，dt，跨越了好几年。数据被分组，因此每个组具有不同的年份长度。我只想保留每组的前三年。我怎样才能用data.table做到这一点？下面是一些要测试的数据：

dates <- c(seq(as.Date('2010-01-03'),as.Date('2019-12-31'),by = 1),
           seq(as.Date('2013-01-02'),as.Date('2018-12-31'),by = 1),
           seq(as.Date('2015-01-02'),as.Date('2020-07-31'),by = 1))

set.seed(1995)
value <- rnorm(length(dates), mean = 100, sd = 50)

IDs <- c(rep(c("ACG"),length.out = length(seq(as.Date('2010-01-03'),as.Date('2019-12-31'),by = 1))),
         rep(c("MKD"),length.out = length(seq(as.Date('2013-01-02'),as.Date('2018-12-31'),by = 1))),
         rep(c("ZED"),length.out = length(seq(as.Date('2015-01-02'),as.Date('2020-07-31'),by = 1)))
)

dt <- data.table(Date = dates,
                 Value = value,
                 ID = IDs
                )
dt

            Date     Value  ID
   1: 2010-01-03 153.03816 ACG
   2: 2010-01-04  83.22491 ACG
   3: 2010-01-05 107.26521 ACG
   4: 2010-01-06 119.70395 ACG
   5: 2010-01-07 183.24604 ACG
  ---                         
7874: 2020-07-27 184.45801 ZED
7875: 2020-07-28  91.53373 ZED
7876: 2020-07-29  67.42443 ZED
7877: 2020-07-30 125.62496 ZED
7878: 2020-07-31  89.02373 ZED

最终的data.table应该只有前三年的数据，如下所示：

finalDT <- dt[c(1:1094,3651:4744,5841:6935),]
finalDT

            Date     Value  ID
   1: 2010-01-03 153.03816 ACG
   2: 2010-01-04  83.22491 ACG
   3: 2010-01-05 107.26521 ACG
   4: 2010-01-06 119.70395 ACG
   5: 2010-01-07 183.24604 ACG
  ---                         
3279: 2017-12-27 102.10622 ZED
3280: 2017-12-28  94.97718 ZED
3281: 2017-12-29 131.47358 ZED
3282: 2017-12-30 112.83836 ZED
3283: 2017-12-31 184.54966 ZED

我使用的方法在较小的数据集上工作得很好，但我有超过100个I，其中一些具有20年的数据。我需要在data.table中使用一种编程方法

Answer 1

dt[, .SD[year(Date) %in% unique(year(Date))[1:3]], by = ID]

或

dt[, .SD[year(Date) <= unique(year(Date))[3]], by = ID]

请确保先按日期排序。