按组查找期间间隔内的日期
Find dates within a period interval by group
我有一个包含许多 ID、begin 和 end 日期的面板。 begin 到 end 日期创建 interval 时间。
id begin end interval overlap
1: 1 2010-01-31 2011-06-30 2009-08-04 UTC--2011-12-27 UTC TRUE
2: 1 2011-01-31 2012-06-30 2010-08-04 UTC--2012-12-27 UTC TRUE
3: 1 2012-01-31 2013-06-30 2011-08-04 UTC--2013-12-27 UTC TRUE
4: 1 2013-01-31 2014-06-30 2012-08-04 UTC--2014-12-27 UTC TRUE
5: 1 2013-02-28 2013-07-31 2012-09-01 UTC--2014-01-27 UTC TRUE
6: 1 2015-02-28 2015-03-31 2014-09-01 UTC--2015-09-27 UTC TRUE
7: 1 2015-06-30 2015-07-31 2015-01-01 UTC--2016-01-27 UTC TRUE
8: 1 2015-09-30 2016-01-31 2015-04-03 UTC--2016-07-29 UTC TRUE
9: 2 2010-01-31 2011-06-30 2009-08-04 UTC--2011-12-27 UTC TRUE
10: 2 2011-01-31 2012-06-30 2010-08-04 UTC--2012-12-27 UTC TRUE
11: 2 2012-01-31 2013-06-30 2011-08-04 UTC--2013-12-27 UTC TRUE
12: 2 2013-01-31 2014-06-30 2012-08-04 UTC--2014-12-27 UTC TRUE
13: 2 2013-02-28 2013-07-31 2012-09-01 UTC--2014-01-27 UTC TRUE
14: 2 2015-02-28 2015-03-31 2014-09-01 UTC--2015-09-27 UTC TRUE
15: 2 2015-06-30 2015-07-31 2015-01-01 UTC--2016-01-27 UTC TRUE
16: 2 2015-09-30 2016-01-31 2015-04-03 UTC--2016-07-29 UTC TRUE
我需要测试，对于每个 ID，任何 begin/end 日期是否包含在另一个 interval（具有相同 ID）中。
例如，id1 begin (2010-01-31) 不包含在 id1 的任何其他时间段中，而不是在第一行中。但是，id1end日期（2011-06-30）包含在第二行的区间（2010-08-04 UTC--2012-12-27 UTC）。
我已经在数据 table 中尝试了 lubridate 间隔和 %within%，但它会产生 TRUE，因为它包含在相应的时间段中。我需要知道它是否包含在同一 ID 的任何其他期间。
customer[begin %within% interval | end %within% interval, overlap := TRUE, by = id]
我已经检查了 data.table 的重叠，但似乎设计用于连接不同的 table 和其他问题只是向量，但 not panels 有间隔。
有什么想法吗？
数据：
structure(list(id = c(1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2,
2, 2, 2), begin = structure(c(14640, 15005, 15370, 15736, 15764,
16494, 16616, 16708, 14640, 15005, 15370, 15736, 15764, 16494,
16616, 16708), class = "Date"), end = structure(c(15155, 15521,
15886, 16251, 15917, 16525, 16647, 16831, 15155, 15521, 15886,
16251, 15917, 16525, 16647, 16831), class = "Date"), interval = structure(c(75600000,
75686400, 75686400, 75600000, 44323200, 33782400, 33782400, 41731200,
75600000, 75686400, 75686400, 75600000, 44323200, 33782400, 33782400,
41731200), start = structure(c(1249344000, 1280880000, 1312416000,
1344038400, 1346457600, 1409529600, 1420070400, 1428019200, 1249344000,
1280880000, 1312416000, 1344038400, 1346457600, 1409529600, 1420070400,
1428019200), tzone = "UTC", class = c("POSIXct", "POSIXt")), tzone = "UTC", class = structure("Interval", package = "lubridate")),
overlap = c(TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE)), .Names = c("id",
"begin", "end", "interval", "overlap"), row.names = c(NA, -16L
), class = c("data.table", "data.frame"), .internal.selfref = <pointer: 0x0000000000140788>)
这是一种方法，使用 lubridate 中的 int_overlaps。我已经定义了从开始日期到结束日期的间隔，尽管在您的数据中它们是不同的 - 也许您可以澄清哪个是正确的。
library(lubridate)
df$interval <- interval(as.POSIXct(df$begin),as.POSIXct(df$end))
df <- df[order(df$id),] #needs to be sorted by id for next stage to work
df$overlap <- unlist(tapply(df$interval, #loop through intervals
df$id, #grouped by id
function(x) rowSums(outer(x,x,int_overlaps))>1))
#check if more than one overlap in subset for that id
df
id begin end interval overlap
1 1 2010-01-31 2011-06-30 2010-01-31 UTC--2011-06-30 UTC TRUE
2 1 2011-01-31 2012-06-30 2011-01-31 UTC--2012-06-30 UTC TRUE
3 1 2012-01-31 2013-06-30 2012-01-31 UTC--2013-06-30 UTC TRUE
4 1 2013-01-31 2014-06-30 2013-01-31 UTC--2014-06-30 UTC TRUE
5 1 2013-02-28 2013-07-31 2013-02-28 UTC--2013-07-31 UTC TRUE
6 1 2015-02-28 2015-03-31 2015-02-28 UTC--2015-03-31 UTC FALSE
7 1 2015-06-30 2015-07-31 2015-06-30 UTC--2015-07-31 UTC FALSE
8 1 2015-09-30 2016-01-31 2015-09-30 UTC--2016-01-31 UTC FALSE
9 2 2010-01-31 2011-06-30 2010-01-31 UTC--2011-06-30 UTC TRUE
10 2 2011-01-31 2012-06-30 2011-01-31 UTC--2012-06-30 UTC TRUE
11 2 2012-01-31 2013-06-30 2012-01-31 UTC--2013-06-30 UTC TRUE
12 2 2013-01-31 2014-06-30 2013-01-31 UTC--2014-06-30 UTC TRUE
13 2 2013-02-28 2013-07-31 2013-02-28 UTC--2013-07-31 UTC TRUE
14 2 2015-02-28 2015-03-31 2015-02-28 UTC--2015-03-31 UTC FALSE
15 2 2015-06-30 2015-07-31 2015-06-30 UTC--2015-07-31 UTC FALSE
16 2 2015-09-30 2016-01-31 2015-09-30 UTC--2016-01-31 UTC FALSE
我有一个包含许多 ID、begin 和 end 日期的面板。 begin 到 end 日期创建 interval 时间。
id begin end interval overlap
1: 1 2010-01-31 2011-06-30 2009-08-04 UTC--2011-12-27 UTC TRUE
2: 1 2011-01-31 2012-06-30 2010-08-04 UTC--2012-12-27 UTC TRUE
3: 1 2012-01-31 2013-06-30 2011-08-04 UTC--2013-12-27 UTC TRUE
4: 1 2013-01-31 2014-06-30 2012-08-04 UTC--2014-12-27 UTC TRUE
5: 1 2013-02-28 2013-07-31 2012-09-01 UTC--2014-01-27 UTC TRUE
6: 1 2015-02-28 2015-03-31 2014-09-01 UTC--2015-09-27 UTC TRUE
7: 1 2015-06-30 2015-07-31 2015-01-01 UTC--2016-01-27 UTC TRUE
8: 1 2015-09-30 2016-01-31 2015-04-03 UTC--2016-07-29 UTC TRUE
9: 2 2010-01-31 2011-06-30 2009-08-04 UTC--2011-12-27 UTC TRUE
10: 2 2011-01-31 2012-06-30 2010-08-04 UTC--2012-12-27 UTC TRUE
11: 2 2012-01-31 2013-06-30 2011-08-04 UTC--2013-12-27 UTC TRUE
12: 2 2013-01-31 2014-06-30 2012-08-04 UTC--2014-12-27 UTC TRUE
13: 2 2013-02-28 2013-07-31 2012-09-01 UTC--2014-01-27 UTC TRUE
14: 2 2015-02-28 2015-03-31 2014-09-01 UTC--2015-09-27 UTC TRUE
15: 2 2015-06-30 2015-07-31 2015-01-01 UTC--2016-01-27 UTC TRUE
16: 2 2015-09-30 2016-01-31 2015-04-03 UTC--2016-07-29 UTC TRUE
我需要测试，对于每个 ID，任何 begin/end 日期是否包含在另一个 interval（具有相同 ID）中。
例如，id1 begin (2010-01-31) 不包含在 id1 的任何其他时间段中，而不是在第一行中。但是，id1end日期（2011-06-30）包含在第二行的区间（2010-08-04 UTC--2012-12-27 UTC）。
我已经在数据 table 中尝试了 lubridate 间隔和 %within%，但它会产生 TRUE，因为它包含在相应的时间段中。我需要知道它是否包含在同一 ID 的任何其他期间。
customer[begin %within% interval | end %within% interval, overlap := TRUE, by = id]
我已经检查了 data.table 的重叠，但似乎设计用于连接不同的 table 和其他问题只是向量，但 not panels 有间隔。
有什么想法吗？
数据：
structure(list(id = c(1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2,
2, 2, 2), begin = structure(c(14640, 15005, 15370, 15736, 15764,
16494, 16616, 16708, 14640, 15005, 15370, 15736, 15764, 16494,
16616, 16708), class = "Date"), end = structure(c(15155, 15521,
15886, 16251, 15917, 16525, 16647, 16831, 15155, 15521, 15886,
16251, 15917, 16525, 16647, 16831), class = "Date"), interval = structure(c(75600000,
75686400, 75686400, 75600000, 44323200, 33782400, 33782400, 41731200,
75600000, 75686400, 75686400, 75600000, 44323200, 33782400, 33782400,
41731200), start = structure(c(1249344000, 1280880000, 1312416000,
1344038400, 1346457600, 1409529600, 1420070400, 1428019200, 1249344000,
1280880000, 1312416000, 1344038400, 1346457600, 1409529600, 1420070400,
1428019200), tzone = "UTC", class = c("POSIXct", "POSIXt")), tzone = "UTC", class = structure("Interval", package = "lubridate")),
overlap = c(TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,
TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE)), .Names = c("id",
"begin", "end", "interval", "overlap"), row.names = c(NA, -16L
), class = c("data.table", "data.frame"), .internal.selfref = <pointer: 0x0000000000140788>)
这是一种方法，使用 lubridate 中的 int_overlaps。我已经定义了从开始日期到结束日期的间隔，尽管在您的数据中它们是不同的 - 也许您可以澄清哪个是正确的。
library(lubridate)
df$interval <- interval(as.POSIXct(df$begin),as.POSIXct(df$end))
df <- df[order(df$id),] #needs to be sorted by id for next stage to work
df$overlap <- unlist(tapply(df$interval, #loop through intervals
df$id, #grouped by id
function(x) rowSums(outer(x,x,int_overlaps))>1))
#check if more than one overlap in subset for that id
df
id begin end interval overlap
1 1 2010-01-31 2011-06-30 2010-01-31 UTC--2011-06-30 UTC TRUE
2 1 2011-01-31 2012-06-30 2011-01-31 UTC--2012-06-30 UTC TRUE
3 1 2012-01-31 2013-06-30 2012-01-31 UTC--2013-06-30 UTC TRUE
4 1 2013-01-31 2014-06-30 2013-01-31 UTC--2014-06-30 UTC TRUE
5 1 2013-02-28 2013-07-31 2013-02-28 UTC--2013-07-31 UTC TRUE
6 1 2015-02-28 2015-03-31 2015-02-28 UTC--2015-03-31 UTC FALSE
7 1 2015-06-30 2015-07-31 2015-06-30 UTC--2015-07-31 UTC FALSE
8 1 2015-09-30 2016-01-31 2015-09-30 UTC--2016-01-31 UTC FALSE
9 2 2010-01-31 2011-06-30 2010-01-31 UTC--2011-06-30 UTC TRUE
10 2 2011-01-31 2012-06-30 2011-01-31 UTC--2012-06-30 UTC TRUE
11 2 2012-01-31 2013-06-30 2012-01-31 UTC--2013-06-30 UTC TRUE
12 2 2013-01-31 2014-06-30 2013-01-31 UTC--2014-06-30 UTC TRUE
13 2 2013-02-28 2013-07-31 2013-02-28 UTC--2013-07-31 UTC TRUE
14 2 2015-02-28 2015-03-31 2015-02-28 UTC--2015-03-31 UTC FALSE
15 2 2015-06-30 2015-07-31 2015-06-30 UTC--2015-07-31 UTC FALSE
16 2 2015-09-30 2016-01-31 2015-09-30 UTC--2016-01-31 UTC FALSE