按组查找下一个日期

Question

我有一些这样的数据:

sample.data <- rbind(data.table(start.date=seq(from=as.Date("2010-01-01"), to=as.Date("2014-12-01"), by="quarter"),
                 Group=c("A","B","C","D"), rnorm(20, 5)),
                 data.table(start.date=seq(from=as.Date("2010-01-01"), to=as.Date("2014-12-01"), by="quarter"),
                 Group=c("A","B","C","D"), rnorm(20, 3))
                 )

我想创建一个end.date等于start.date每个组的下一个最早值的列.

因此,例如,第一个start.date为Group==A是2010-01-01.下一个最早start.date的Group==A是2011-01-01.所以当按以下顺序排序时,最终结果应如下所示Group:

                start.date Group   end.date
                2010-01-01     A 2011-01-01
                2010-01-01     A 2011-01-01
                2011-01-01     A 2012-01-01
                2011-01-01     A 2012-01-01
                2012-01-01     A 2013-01-01
                2012-01-01     A 2013-01-01
                2013-01-01     A 2014-01-01
                2013-01-01     A 2014-01-01
                2014-01-01     A         NA
                2014-01-01     A         NA
                2010-04-01     B 2011-04-01
                2010-04-01     B 2011-04-01
                2011-04-01     B 2012-04-01
                2011-04-01     B 2012-04-01

等等.理想情况下,我想通过引用来做到这一点,比如

sample.data[, end.date := EXPRESSION]

但我不知道从哪里开始.谢谢你的帮助.

Answer 1

可以,然后呢:

events = unique(sample.data[ , .(Group, start.date) ])
events[, next.date := shift(start.date, type="lead"), by=Group]

sample.data[events, on=c("Group", "start.date"), end.date := next.date ]

在我看来,OP应该有一个类似的表events,与数据库设计/ 整洁的数据保持一致.结果看起来像

> sample.data[ order(Group, start.date) ]

    start.date Group   end.date
 1: 2010-01-01     A 2011-01-01
 2: 2010-01-01     A 2011-01-01
 3: 2011-01-01     A 2012-01-01
 4: 2011-01-01     A 2012-01-01
 5: 2012-01-01     A 2013-01-01
 6: 2012-01-01     A 2013-01-01
 7: 2013-01-01     A 2014-01-01
 8: 2013-01-01     A 2014-01-01
 9: 2014-01-01     A       <NA>
10: 2014-01-01     A       <NA>
11: 2010-04-01     B 2011-04-01
12: 2010-04-01     B 2011-04-01
...