r - 查找每行最接近特定值的时间

Question

假设我有一个数据表:

dt <- data.table(
        datetime = seq(as.POSIXct("2016-01-01 00:00:00"),as.POSIXct("2016-01-01 10:00:00"), by = "1 hour"),
        ObType = c("A","A","B","B","B","B","A","A","B","A","A")
)

dt
                   datetime ObType
     1: 2016-01-01 00:00:00      A
     2: 2016-01-01 01:00:00      A
     3: 2016-01-01 02:00:00      B
     4: 2016-01-01 03:00:00      B
     5: 2016-01-01 04:00:00      B
     6: 2016-01-01 05:00:00      B
     7: 2016-01-01 06:00:00      A
     8: 2016-01-01 07:00:00      A
     9: 2016-01-01 08:00:00      B
    10: 2016-01-01 09:00:00      A
    11: 2016-01-01 10:00:00      A

我需要做的是在 ObType 为“B”的任何地方，我都需要找到到任一侧最近的 ObType“A”的时间。所以结果应该是这样的(以小时为单位):

               datetime ObType timeLag timeLead
 1: 2016-01-01 00:00:00      A      NA       NA
 2: 2016-01-01 01:00:00      A      NA       NA
 3: 2016-01-01 02:00:00      B       1        4
 4: 2016-01-01 03:00:00      B       2        3
 5: 2016-01-01 04:00:00      B       3        2
 6: 2016-01-01 05:00:00      B       4        1
 7: 2016-01-01 06:00:00      A      NA       NA
 8: 2016-01-01 07:00:00      A      NA       NA
 9: 2016-01-01 08:00:00      B       1        1
10: 2016-01-01 09:00:00      A      NA       NA
11: 2016-01-01 10:00:00      A      NA       NA

我通常使用 data.table，但非 data.table 解决方案也很好。

谢谢!

赖斯

Answer 1

我暗示使用 roll= 的方法:

X = dt[ObType=="A"]
X
              datetime ObType
1: 2016-01-01 00:00:00      A
2: 2016-01-01 01:00:00      A
3: 2016-01-01 06:00:00      A
4: 2016-01-01 07:00:00      A
5: 2016-01-01 09:00:00      A
6: 2016-01-01 10:00:00      A

dt[ObType=="B", Lag:=X[.SD,on="datetime",roll=Inf,i.datetime-x.datetime]]
dt[ObType=="B", Lead:=X[.SD,on="datetime",roll=-Inf,x.datetime-i.datetime]]
dt[ObType=="B", Nearest:=X[.SD,on="datetime",roll="nearest",x.datetime-i.datetime]]
dt
               datetime ObType      Lag     Lead     Nearest
 1: 2016-01-01 00:00:00      A NA hours NA hours    NA hours
 2: 2016-01-01 01:00:00      A NA hours NA hours    NA hours
 3: 2016-01-01 02:00:00      B  1 hours  4 hours    -1 hours
 4: 2016-01-01 03:00:00      B  2 hours  3 hours    -2 hours
 5: 2016-01-01 04:00:00      B  3 hours  2 hours     2 hours
 6: 2016-01-01 05:00:00      B  4 hours  1 hours     1 hours
 7: 2016-01-01 06:00:00      A NA hours NA hours    NA hours
 8: 2016-01-01 07:00:00      A NA hours NA hours    NA hours
 9: 2016-01-01 08:00:00      B  1 hours  1 hours    -1 hours
10: 2016-01-01 09:00:00      A NA hours NA hours    NA hours
11: 2016-01-01 10:00:00      A NA hours NA hours    NA hours

roll=的优势之一是您可以通过更改 Inf 来应用陈旧限制。到您希望加入的时间限制。这是限制适用的时间差，而不是行数。 Inf只是意味着不限制。 roll=符号表示是向前看还是向后看(领先或滞后)。

另一个优点是roll=很快。