具有重叠时间间隔的 R 组 ID

Question

我有一个包含多个组的大型数据集，其中包含具有开始和停止日期时间的 ID 数据集。我要做的是在每个组内确定子组发生的位置。当两个 ID 与其 START & END 日期时间列重叠时，组中的子组将是一个子组。下面是在 R 中为一组创建示例数据集的脚本。我想要做的是在每个组中创建一个名为“Grp”的列，将这些子组与重叠的 START 和 END 日期时间分组。

我有什么...

structure(list(ID = c(1,2,3,4), START = structure(c(1490904000, 1490918400, 
1508363100, 1508379300), tzone = "UTC", class = c("POSIXct", 
"POSIXt")), END = structure(c(1492050600, 1492247700, 
1509062400, 1509031800), tzone = "UTC", class = c("POSIXct", 
"POSIXt"))), class = "data.frame", row.names = c(NA, -4L), .Names = c("ID","START", 
"END")) 

我想要的是……

structure(list(ID = c(1,2,3,4), START = structure(c(1490904000, 1508379300, 
1508363100, 1490918400), tzone = "UTC", class = c("POSIXct", 
"POSIXt")), END = structure(c(1492050600, 1509031800, 
1509062400, 1492247700), tzone = "UTC", class = c("POSIXct", 
"POSIXt")), Grp = c(1,2,2,1)), class = "data.frame", row.names = c(NA, -4L), .Names = c("ID","START", 
"END","Grp"))

我尝试使用 lubridate 的间隔，并以这种方式找到重叠，但没有运气。任何帮助将不胜感激。

Answer 1

按START排序后，新组的条件是上一行的END小于下一行的START：

head(df1$END, -1) < tail(df1$START,-1)

df1 <- structure(list(ID = c(1,2,3,4), START = structure(c(1490904000, 1490918400, 
1508363100, 1508379300), tzone = "UTC", class = c("POSIXct", 
"POSIXt")), END = structure(c(1492050600, 1492247700, 
1509062400, 1509031800), tzone = "UTC", class = c("POSIXct", 
"POSIXt"))), class = "data.frame", row.names = c(NA, -4L), .Names = c("ID","START", 
"END"))

df1
  ID               START                 END
1  1 2017-03-30 20:00:00 2017-04-13 02:30:00
2  2 2017-03-31 00:00:00 2017-04-15 09:15:00
3  3 2017-10-18 21:45:00 2017-10-27 00:00:00
4  4 2017-10-19 02:15:00 2017-10-26 15:30:00


df1a <- df1[ order(df1$START), ]
df1a$grp <- cumsum( c( 1, head(df1$END, -1) < tail(df1$START,-1) ))
 df1a
#---------------
  ID               START                 END grp
1  1 2017-03-30 20:00:00 2017-04-13 02:30:00   1
2  2 2017-03-31 00:00:00 2017-04-15 09:15:00   1
3  3 2017-10-18 21:45:00 2017-10-27 00:00:00   2
4  4 2017-10-19 02:15:00 2017-10-26 15:30:00   2

这是一个函数，可以回答我对以下评论的回复的第一部分：

grp_overlaps <- function(endings, beginnings){ 
                   cumsum(c( 1, head(endings, -1) < tail(beginnings, -1) )) }