根据 R 中的多个条件（时间差和因子）按组折叠行
Collapse rows by group based on multiple conditions (time difference and factor) in R
我希望根据时间戳 and/or 之间指定的时间差（即 60 分钟）按组折叠数据行，直到在数据中满足特定条件。这是我正在使用的模拟数据框：-
UserId<-c("2203af12ce3e", "2203af12ce3e", "2203af12ce3e", "2203af12ce3e",
"3b9c32d4c700", "3b9c32d4c700", "3b9c32d4c700", "3b9c32d4c700",
"3b9c32d4c700", "68b25fd3ca78", "68b25fd3ca78", "68b25fd3ca78",
"68b25fd3ca78", "68b25fd3ca78", "fbbd0e13e61b", "fbbd0e13e61b",
"fbbd0e13e61b", "808dcbe0cad2", "6f2020841f9e", "faf849c35400",
"02002044e512", "02002044e512", "02002044e512", "02002044e512",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791")
OrigTime<-structure(c(1546313039, 1546313070, 1546313116, 1546344887, 1546366783,
1546371206, 1546378029, 1546380713, 1546381727, 1546317095, 1546317335,
1546319551, 1546347453, 1546355351, 1546381815, 1546381844, 1546381873,
1546355462, 1546370527, 1546354015, 1546310854, 1546311154, 1546311218,
1546311772, 1546308212, 1546308230, 1546308248, 1546308268, 1546308298,
1546308317, 1546330603, 1546330620, 1546330637, 1546330650, 1546330676,
1546334052, 1546334129, 1546334523, 1546334867, 1546334902), class = c("POSIXct",
"POSIXt"), tzone = "UTC")
LastTime<-structure(c(1546313039, 1546313070, 1546313448, 1546344887, 1546366783,
1546371206, 1546378029, 1546380713, 1546381727, 1546317095, 1546318123,
1546319551, 1546347453, 1546355351, 1546381815, 1546381844, 1546381873,
1546355462, 1546370527, 1546354015, 1546311063, 1546311154, 1546311746,
1546313128, 1546308212, 1546308230, 1546308248, 1546308268, 1546308298,
1546310346, 1546330603, 1546330620, 1546330637, 1546330650, 1546330676,
1546334052, 1546334129, 1546334523, 1546334867, 1546334902), class = c("POSIXct",
"POSIXt"), tzone = "UTC")
calls<-c(1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1,
4, 1, 3, 4, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
Status<-c("Engaged", "Engaged", "Abandoned", "Abandoned", "Answered",
"Answered", "Answered", "Answered", "Answered", "Engaged", "Engaged",
"Engaged", "Answered", "Answered", "Engaged", "Engaged", "Answered",
"Answered", "Answered", "Answered", "Engaged", "Engaged", "Engaged",
"Abandoned", "Engaged", "Engaged", "Engaged", "Engaged", "Engaged",
"Answered", "Engaged", "Engaged", "Engaged", "Engaged", "Answered",
"Answered", "Answered", "Answered", "Answered", "Answered")
Successful<-c("N", "N", "N", "N", "Y", "Y", "Y", "Y", "Y", "N", "N", "N",
"Y", "Y", "N", "N", "Y", "Y", "Y", "Y", "N", "N", "N", "N", "N",
"N", "N", "N", "N", "Y", "N", "N", "N", "N", "Y", "Y", "Y", "Y",
"Y", "Y")
df<-as.data.frame(cbind(UserId,OrigTime,LastTime,calls,Status,Successful))
df$OrigTime<-as.numeric(df$OrigTime)
df$OrigTime<-as.POSIXct(df$OrigTime, origin = "1970-01-01 00:00:00")
df$LastTime<-as.numeric(df$LastTime)
df$LastTime<-as.POSIXct(df$LastTime, origin = "1970-01-01 00:00:00")
下面是我正在努力实现的一些示例：-
示例 1
这里是UserId = "2203af12ce3e",在df:-
之前
df%>%filter(UserId=="2203af12ce3e")
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 03:23:59 2019-01-01 03:23:59 1 Engaged N 2203af12ce3e
2 2019-01-01 03:24:30 2019-01-01 03:24:30 1 Engaged N 2203af12ce3e
3 2019-01-01 03:25:16 2019-01-01 03:30:48 3 Abandoned N 2203af12ce3e
4 2019-01-01 12:14:47 2019-01-01 12:14:47 1 Abandoned N 2203af12ce3e
我想折叠第一个 LastTime 和随后的 OrigTime 之间 60 分钟内的行，同时统计 calls 并显示最后一个 [=18] =].所以在这种情况下，我会将第 1-3 行折叠成一行，计算调用次数，最后状态为“已放弃”，如下所示：-
结果
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 03:23:59 2019-01-01 03:30:48 5 Abandoned N 2203af12ce3e
2 2019-01-01 12:14:47 2019-01-01 12:14:47 1 Abandoned N 2203af12ce3e
请注意，Before 段中第 1 行中的 OrigTime 和第 3 行中的 LastTime 现在在一行中，来自第 1-3 行的调用被统计并从第 3 行中获取最后一个 Status，因为第 1- 3 都发生在 60 分钟内 window。第 4 行保留原样，因为它的 OrigTime 比上一个 LastTime 晚了 >60 分钟。
我想再举一个例子来说明我想做什么。
例2
如果在“已回答”因素出现之前有很多“已放弃”和“参与”因素，并且所有因素都发生在 60 分钟内 window，那么我希望这些行与最后一个一起折叠“已回答”的状态。这是示例 2：-
之前
df%>%filter(UserId=="ff10b8560791")
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 02:03:32 2019-01-01 02:03:32 1 Engaged N ff10b8560791
2 2019-01-01 02:03:50 2019-01-01 02:03:50 1 Engaged N ff10b8560791
3 2019-01-01 02:04:08 2019-01-01 02:04:08 1 Engaged N ff10b8560791
4 2019-01-01 02:04:28 2019-01-01 02:04:28 1 Engaged N ff10b8560791
5 2019-01-01 02:04:58 2019-01-01 02:04:58 1 Engaged N ff10b8560791
6 2019-01-01 02:05:17 2019-01-01 02:39:06 3 Answered Y ff10b8560791
7 2019-01-01 08:16:43 2019-01-01 08:16:43 1 Engaged N ff10b8560791
8 2019-01-01 08:17:00 2019-01-01 08:17:00 1 Engaged N ff10b8560791
9 2019-01-01 08:17:17 2019-01-01 08:17:17 1 Engaged N ff10b8560791
10 2019-01-01 08:17:30 2019-01-01 08:17:30 1 Engaged N ff10b8560791
11 2019-01-01 08:17:56 2019-01-01 08:17:56 1 Answered Y ff10b8560791
12 2019-01-01 09:14:12 2019-01-01 09:14:12 1 Answered Y ff10b8560791
13 2019-01-01 09:15:29 2019-01-01 09:15:29 1 Answered Y ff10b8560791
14 2019-01-01 09:22:03 2019-01-01 09:22:03 1 Answered Y ff10b8560791
15 2019-01-01 09:27:47 2019-01-01 09:27:47 1 Answered Y ff10b8560791
16 2019-01-01 09:28:22 2019-01-01 09:28:22 1 Answered Y ff10b8560791
与前面的示例一样，我想折叠在 OrigTime 和之前的 LastTime 的 60 分钟内发生的行，统计调用列并给出该行的最后状态。但是，这里有一个额外的条件：如果该行具有“已回答”状态并且之前的行都发生在该“已回答”行的 60 分钟内，那么我想停在那里并统计之前的行（如果它们发生在 60 分钟内）分钟 window，最终状态为“已回答”。这是此示例的结果：-
结果
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 02:03:32 2019-01-01 02:39:06 8 Answered Y ff10b8560791
2 2019-01-01 08:16:43 2019-01-01 08:17:56 5 Answered Y ff10b8560791
3 2019-01-01 09:14:12 2019-01-01 09:14:12 1 Answered Y ff10b8560791
4 2019-01-01 09:15:29 2019-01-01 09:15:29 1 Answered Y ff10b8560791
5 2019-01-01 09:22:03 2019-01-01 09:22:03 1 Answered Y ff10b8560791
6 2019-01-01 09:27:47 2019-01-01 09:27:47 1 Answered Y ff10b8560791
7 2019-01-01 09:28:22 2019-01-01 09:28:22 1 Answered Y ff10b8560791
所以在这种情况下，Before 段中的第 1-6 行已折叠，因为它们都在 60 分钟内发生，但在状态为“已回答”时停止；对于第 7-11 行再次发生，对于具有“已回答”状态的其余行，这保持不变。我希望我提供的结果部分能够清楚地说明我正在尝试做什么。任何帮助将不胜感激:)
我们为每个人创建一个新组：
UserId- 如果
status = 'Answered'
- 60 分钟间隔
对于创建的每个组，我们 select
- 第一个
OrigTime
- 最后一个
LastTime
sum 共 Calls- 最后
Status
- 最后
Successful.
library(dplyr)
df %>%
mutate(grp1 = lag(cumsum(Status == 'Answered'), default = 0)) %>%
group_by(UserId) %>%
mutate(grp2 = floor(as.numeric(difftime(LastTime,
first(OrigTime), units = 'hours')))) %>%
group_by(UserId, grp1, grp2) %>%
summarise(OrigTime = first(OrigTime),
LastTime = last(LastTime),
calls = sum(calls),
Status = last(Status),
Successful = last(Successful)) %>%
ungroup -> result
检查结果：
result %>% filter(UserId == '2203af12ce3e')
# UserId grp1 grp2 OrigTime LastTime calls Status Successful
# <chr> <dbl> <dbl> <dttm> <dttm> <dbl> <chr> <chr>
#1 2203af12ce3e 0 0 2019-01-01 11:23:59 2019-01-01 11:30:48 5 Abandoned N
#2 2203af12ce3e 0 8 2019-01-01 20:14:47 2019-01-01 20:14:47 1 Abandoned N
result %>% filter(UserId == 'ff10b8560791')
# A tibble: 7 x 8
# UserId grp1 grp2 OrigTime LastTime calls Status Successful
# <chr> <dbl> <dbl> <dttm> <dttm> <dbl> <chr> <chr>
#1 ff10b8560791 11 0 2019-01-01 10:03:32 2019-01-01 10:39:06 8 Answered Y
#2 ff10b8560791 12 6 2019-01-01 16:16:43 2019-01-01 16:17:56 5 Answered Y
#3 ff10b8560791 13 7 2019-01-01 17:14:12 2019-01-01 17:14:12 1 Answered Y
#4 ff10b8560791 14 7 2019-01-01 17:15:29 2019-01-01 17:15:29 1 Answered Y
#5 ff10b8560791 15 7 2019-01-01 17:22:03 2019-01-01 17:22:03 1 Answered Y
#6 ff10b8560791 16 7 2019-01-01 17:27:47 2019-01-01 17:27:47 1 Answered Y
#7 ff10b8560791 17 7 2019-01-01 17:28:22 2019-01-01 17:28:22 1 Answered Y
由于我们所在的时区不同，时间值不同。
要构造数据框，请不要 cbind（因为它将数据转换为矩阵）直接使用 data.frame。
df<- data.frame(UserId,OrigTime,LastTime,calls,Status,Successful)
我希望根据时间戳 and/or 之间指定的时间差（即 60 分钟）按组折叠数据行，直到在数据中满足特定条件。这是我正在使用的模拟数据框：-
UserId<-c("2203af12ce3e", "2203af12ce3e", "2203af12ce3e", "2203af12ce3e",
"3b9c32d4c700", "3b9c32d4c700", "3b9c32d4c700", "3b9c32d4c700",
"3b9c32d4c700", "68b25fd3ca78", "68b25fd3ca78", "68b25fd3ca78",
"68b25fd3ca78", "68b25fd3ca78", "fbbd0e13e61b", "fbbd0e13e61b",
"fbbd0e13e61b", "808dcbe0cad2", "6f2020841f9e", "faf849c35400",
"02002044e512", "02002044e512", "02002044e512", "02002044e512",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791",
"ff10b8560791", "ff10b8560791", "ff10b8560791", "ff10b8560791")
OrigTime<-structure(c(1546313039, 1546313070, 1546313116, 1546344887, 1546366783,
1546371206, 1546378029, 1546380713, 1546381727, 1546317095, 1546317335,
1546319551, 1546347453, 1546355351, 1546381815, 1546381844, 1546381873,
1546355462, 1546370527, 1546354015, 1546310854, 1546311154, 1546311218,
1546311772, 1546308212, 1546308230, 1546308248, 1546308268, 1546308298,
1546308317, 1546330603, 1546330620, 1546330637, 1546330650, 1546330676,
1546334052, 1546334129, 1546334523, 1546334867, 1546334902), class = c("POSIXct",
"POSIXt"), tzone = "UTC")
LastTime<-structure(c(1546313039, 1546313070, 1546313448, 1546344887, 1546366783,
1546371206, 1546378029, 1546380713, 1546381727, 1546317095, 1546318123,
1546319551, 1546347453, 1546355351, 1546381815, 1546381844, 1546381873,
1546355462, 1546370527, 1546354015, 1546311063, 1546311154, 1546311746,
1546313128, 1546308212, 1546308230, 1546308248, 1546308268, 1546308298,
1546310346, 1546330603, 1546330620, 1546330637, 1546330650, 1546330676,
1546334052, 1546334129, 1546334523, 1546334867, 1546334902), class = c("POSIXct",
"POSIXt"), tzone = "UTC")
calls<-c(1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1,
4, 1, 3, 4, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
Status<-c("Engaged", "Engaged", "Abandoned", "Abandoned", "Answered",
"Answered", "Answered", "Answered", "Answered", "Engaged", "Engaged",
"Engaged", "Answered", "Answered", "Engaged", "Engaged", "Answered",
"Answered", "Answered", "Answered", "Engaged", "Engaged", "Engaged",
"Abandoned", "Engaged", "Engaged", "Engaged", "Engaged", "Engaged",
"Answered", "Engaged", "Engaged", "Engaged", "Engaged", "Answered",
"Answered", "Answered", "Answered", "Answered", "Answered")
Successful<-c("N", "N", "N", "N", "Y", "Y", "Y", "Y", "Y", "N", "N", "N",
"Y", "Y", "N", "N", "Y", "Y", "Y", "Y", "N", "N", "N", "N", "N",
"N", "N", "N", "N", "Y", "N", "N", "N", "N", "Y", "Y", "Y", "Y",
"Y", "Y")
df<-as.data.frame(cbind(UserId,OrigTime,LastTime,calls,Status,Successful))
df$OrigTime<-as.numeric(df$OrigTime)
df$OrigTime<-as.POSIXct(df$OrigTime, origin = "1970-01-01 00:00:00")
df$LastTime<-as.numeric(df$LastTime)
df$LastTime<-as.POSIXct(df$LastTime, origin = "1970-01-01 00:00:00")
下面是我正在努力实现的一些示例：-
示例 1
这里是UserId = "2203af12ce3e",在df:-
之前
df%>%filter(UserId=="2203af12ce3e")
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 03:23:59 2019-01-01 03:23:59 1 Engaged N 2203af12ce3e
2 2019-01-01 03:24:30 2019-01-01 03:24:30 1 Engaged N 2203af12ce3e
3 2019-01-01 03:25:16 2019-01-01 03:30:48 3 Abandoned N 2203af12ce3e
4 2019-01-01 12:14:47 2019-01-01 12:14:47 1 Abandoned N 2203af12ce3e
我想折叠第一个 LastTime 和随后的 OrigTime 之间 60 分钟内的行，同时统计 calls 并显示最后一个 [=18] =].所以在这种情况下，我会将第 1-3 行折叠成一行，计算调用次数，最后状态为“已放弃”，如下所示：-
结果
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 03:23:59 2019-01-01 03:30:48 5 Abandoned N 2203af12ce3e
2 2019-01-01 12:14:47 2019-01-01 12:14:47 1 Abandoned N 2203af12ce3e
请注意，Before 段中第 1 行中的 OrigTime 和第 3 行中的 LastTime 现在在一行中，来自第 1-3 行的调用被统计并从第 3 行中获取最后一个 Status，因为第 1- 3 都发生在 60 分钟内 window。第 4 行保留原样，因为它的 OrigTime 比上一个 LastTime 晚了 >60 分钟。
我想再举一个例子来说明我想做什么。
例2
如果在“已回答”因素出现之前有很多“已放弃”和“参与”因素，并且所有因素都发生在 60 分钟内 window，那么我希望这些行与最后一个一起折叠“已回答”的状态。这是示例 2：-
之前
df%>%filter(UserId=="ff10b8560791")
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 02:03:32 2019-01-01 02:03:32 1 Engaged N ff10b8560791
2 2019-01-01 02:03:50 2019-01-01 02:03:50 1 Engaged N ff10b8560791
3 2019-01-01 02:04:08 2019-01-01 02:04:08 1 Engaged N ff10b8560791
4 2019-01-01 02:04:28 2019-01-01 02:04:28 1 Engaged N ff10b8560791
5 2019-01-01 02:04:58 2019-01-01 02:04:58 1 Engaged N ff10b8560791
6 2019-01-01 02:05:17 2019-01-01 02:39:06 3 Answered Y ff10b8560791
7 2019-01-01 08:16:43 2019-01-01 08:16:43 1 Engaged N ff10b8560791
8 2019-01-01 08:17:00 2019-01-01 08:17:00 1 Engaged N ff10b8560791
9 2019-01-01 08:17:17 2019-01-01 08:17:17 1 Engaged N ff10b8560791
10 2019-01-01 08:17:30 2019-01-01 08:17:30 1 Engaged N ff10b8560791
11 2019-01-01 08:17:56 2019-01-01 08:17:56 1 Answered Y ff10b8560791
12 2019-01-01 09:14:12 2019-01-01 09:14:12 1 Answered Y ff10b8560791
13 2019-01-01 09:15:29 2019-01-01 09:15:29 1 Answered Y ff10b8560791
14 2019-01-01 09:22:03 2019-01-01 09:22:03 1 Answered Y ff10b8560791
15 2019-01-01 09:27:47 2019-01-01 09:27:47 1 Answered Y ff10b8560791
16 2019-01-01 09:28:22 2019-01-01 09:28:22 1 Answered Y ff10b8560791
与前面的示例一样，我想折叠在 OrigTime 和之前的 LastTime 的 60 分钟内发生的行，统计调用列并给出该行的最后状态。但是，这里有一个额外的条件：如果该行具有“已回答”状态并且之前的行都发生在该“已回答”行的 60 分钟内，那么我想停在那里并统计之前的行（如果它们发生在 60 分钟内）分钟 window，最终状态为“已回答”。这是此示例的结果：-
结果
OrigTime LastTime calls Status Successful UserId
1 2019-01-01 02:03:32 2019-01-01 02:39:06 8 Answered Y ff10b8560791
2 2019-01-01 08:16:43 2019-01-01 08:17:56 5 Answered Y ff10b8560791
3 2019-01-01 09:14:12 2019-01-01 09:14:12 1 Answered Y ff10b8560791
4 2019-01-01 09:15:29 2019-01-01 09:15:29 1 Answered Y ff10b8560791
5 2019-01-01 09:22:03 2019-01-01 09:22:03 1 Answered Y ff10b8560791
6 2019-01-01 09:27:47 2019-01-01 09:27:47 1 Answered Y ff10b8560791
7 2019-01-01 09:28:22 2019-01-01 09:28:22 1 Answered Y ff10b8560791
所以在这种情况下，Before 段中的第 1-6 行已折叠，因为它们都在 60 分钟内发生，但在状态为“已回答”时停止；对于第 7-11 行再次发生，对于具有“已回答”状态的其余行，这保持不变。我希望我提供的结果部分能够清楚地说明我正在尝试做什么。任何帮助将不胜感激:)
我们为每个人创建一个新组：
UserId- 如果
status = 'Answered' - 60 分钟间隔
对于创建的每个组，我们 select
- 第一个
OrigTime - 最后一个
LastTime sum共Calls- 最后
Status - 最后
Successful.
library(dplyr)
df %>%
mutate(grp1 = lag(cumsum(Status == 'Answered'), default = 0)) %>%
group_by(UserId) %>%
mutate(grp2 = floor(as.numeric(difftime(LastTime,
first(OrigTime), units = 'hours')))) %>%
group_by(UserId, grp1, grp2) %>%
summarise(OrigTime = first(OrigTime),
LastTime = last(LastTime),
calls = sum(calls),
Status = last(Status),
Successful = last(Successful)) %>%
ungroup -> result
检查结果：
result %>% filter(UserId == '2203af12ce3e')
# UserId grp1 grp2 OrigTime LastTime calls Status Successful
# <chr> <dbl> <dbl> <dttm> <dttm> <dbl> <chr> <chr>
#1 2203af12ce3e 0 0 2019-01-01 11:23:59 2019-01-01 11:30:48 5 Abandoned N
#2 2203af12ce3e 0 8 2019-01-01 20:14:47 2019-01-01 20:14:47 1 Abandoned N
result %>% filter(UserId == 'ff10b8560791')
# A tibble: 7 x 8
# UserId grp1 grp2 OrigTime LastTime calls Status Successful
# <chr> <dbl> <dbl> <dttm> <dttm> <dbl> <chr> <chr>
#1 ff10b8560791 11 0 2019-01-01 10:03:32 2019-01-01 10:39:06 8 Answered Y
#2 ff10b8560791 12 6 2019-01-01 16:16:43 2019-01-01 16:17:56 5 Answered Y
#3 ff10b8560791 13 7 2019-01-01 17:14:12 2019-01-01 17:14:12 1 Answered Y
#4 ff10b8560791 14 7 2019-01-01 17:15:29 2019-01-01 17:15:29 1 Answered Y
#5 ff10b8560791 15 7 2019-01-01 17:22:03 2019-01-01 17:22:03 1 Answered Y
#6 ff10b8560791 16 7 2019-01-01 17:27:47 2019-01-01 17:27:47 1 Answered Y
#7 ff10b8560791 17 7 2019-01-01 17:28:22 2019-01-01 17:28:22 1 Answered Y
由于我们所在的时区不同，时间值不同。
要构造数据框，请不要 cbind（因为它将数据转换为矩阵）直接使用 data.frame。
df<- data.frame(UserId,OrigTime,LastTime,calls,Status,Successful)