根据因子对重叠时间间隔进行分类答案

【问题标题】：Classifying overlapping time intervals based on factor根据因子对重叠时间间隔进行分类
【发布时间】：2020-12-18 16:00:43
【问题描述】：

我正在查看来自制造工厂的生产数据，需要根据一条生产线生产的瓶子大小对所有生产线的当天生产进行分类。基本设置是 FL 5S 生产线仅生产 5l 罐，FL 25 生产 25l 罐，FL 5 可生产 1l、3l 和 5l 罐。我需要一个名为“Recipe”的新列，它应该反映在 FL 5 上生产的所有订单与在 FL 5 上同时生产的订单的所有订单的大小。例如，如果 Size == "001" 在时间间隔 2020-04-28 08:46:56 UTC--2020-04-29 07:49:09 UTC 上Recipe 应该是 001 对于其他行产生的所有订单，其 Started 和 Ended 时间与 1l 订单的这些时间重叠。

数据：

library(tidyverse)
library(lubridate)
Prod_data_ex<-structure(list(Order = c(27380, 27388, 27395, 27381, 27389, 27382, 
                                       27396, 27397, 27393, 27392, 27383, 27384, 27385, 27386, 27398, 
                                       27409, 27410, 27411, 27412, 27416, 27420, 27421, 27417, 27418, 
                                       27432, 27433, 27419, 27413, 27399, 27414, 27415, 27428, 27424, 
                                       27429, 27431), 
                             Prod_line = c("FL 5S", "FL 5", "FL 25", "FL 5S", 
                                           "FL 5", "FL 5S", "FL 25", "FL 5", "FL 5", "FL 5", "FL 5S", "FL 5S", 
                                           "FL 5S", "FL 5S", "FL 5", "FL 5S", "FL 5S", "FL 5S", "FL 5S", 
                                           "FL 25", "FL 5S", "FL 5S", "FL 25", "FL 25", "FL 5S", "FL 5S", 
                                          "FL 25", "FL 5S", "FL 5", "FL 5S", "FL 5S", "FL 5S", "FL 25", 
                                          "FL 5S", "FL 5"), 
                             Produced = c(5400, 6373, 1440, 6372, 864, 5400, 
                                          288, 1080, 864, 5402, 3240, 864, 5293, 2700, 11547, 4427, 3672, 
                                          864, 2651, 96, 648, 1620, 96, 480, 2160, 1363, 480, 4320, 11528, 
                                          4320, 864, 12745, 600, 3782, 6373), 
                             Size = c("005", "005", "025","005", "005", "005", "025", "005", "005", "005", "005", "005", 
                                      "005", "005", "001", "005", "005", "005", "005", "025", "005", 
                                      "005", "025", "025", "005", "005", "025", "005", "001", "005", 
                                      "005", "005", "025", "005", "005"), 
                             Started = structure(c(1587969512, 1587970869, 1587972010, 1587984935, 1587998820, 1587999421, 1588003819, 
                                                  1588004233, 1588008484, 1588013628, 1588014019, 1588022405, 1588024038, 
                                                  1588062901, 1588063616, 1588070291, 1588076547, 1588083066, 1588085684, 
                                                  1588090216, 1588093080, 1588094491, 1588094881, 1588100218, 1588101215, 
                                                  1588105405, 1588105729, 1588110352, 1588146549, 1588147618, 1588155560, 
                                                  1588157595, 1588172423, 1588179841, 1588180603), tzone = "UTC", class = c("POSIXct", 
                                                  "POSIXt")), Ended = structure(c(1587984935, 1587998820, 1588003819, 
                                                  1587999421, 1588004233, 1588014019, 1588090216, 1588008484, 1588013628, 
                                                  1588063616, 1588022405, 1588024038, 1588062901, 1588070291, 1588146549, 
                                                  1588076547, 1588083066, 1588085684, 1588093080, 1588094881, 1588094491, 
                                                  1588101215, 1588100218, 1588105729, 1588105405, 1588110352, 1588172423, 
                                                  1588147618, 1588180603, 1588155560, 1588157595, 1588179841, 1588188834, 
                                                  1588191295, 1588241708), tzone = "UTC", class = c("POSIXct", 
                                                  "POSIXt"))), row.names = c(NA, -35L), class = c("tbl_df", "tbl", "data.frame"))

这是我最接近工作的代码，但仍然不适用于其他行的订单。

Prod_data_recpies<- Prod_data_ex%>%
  mutate(interval= interval(Started, Ended))%>%
  mutate(recipe= ifelse(Size=="001", map_int(interval,~ any(int_overlaps(.x, interval[Size=="001"]))),
                        ifelse(Size=="003", map_int(interval,~ any(int_overlaps(.x, interval[Size=="003"]))), "005")))

谢谢！

【问题讨论】：

标签： r dplyr lubridate

【解决方案1】：

我不确定您想如何处理多个 FL 5 订单的重叠和不存在的重叠，但您可以修改以下方法以满足您的需求。我使用data.table 包中的foverlaps() 来识别重叠。如果您想指定间隔重叠（例如完全在、开始或结束重叠），您可以使用type 参数。

library(data.table)
dt <- as.data.table(Prod_data_ex)
dt[, Ended := Ended - 1] # prevent overlaps of consecutive time intervals
setkeyv(dt, c("Started", "Ended"))
dt[, Recipe := foverlaps(dt, dt[Prod_line == "FL 5"], which = TRUE)[,
  list(list(unique(yid[yid != xid]))), by = xid][, -1]]
dt$Recipe <- sapply(dt$Recipe, function(x) {
  paste(unique(dt[Prod_line == "FL 5"][x]$Size), collapse = ",")
})
dt[, Ended := Ended + 1][] # add back the second

结果：

#>     Order Prod_line Produced Size             Started               Ended
#>  1: 27380     FL 5S     5400  005 2020-04-27 06:38:32 2020-04-27 10:55:35
#>  2: 27388      FL 5     6373  005 2020-04-27 07:01:09 2020-04-27 14:47:00
#>  3: 27395     FL 25     1440  025 2020-04-27 07:20:10 2020-04-27 16:10:19
#>  4: 27381     FL 5S     6372  005 2020-04-27 10:55:35 2020-04-27 14:57:01
#>  5: 27389      FL 5      864  005 2020-04-27 14:47:00 2020-04-27 16:17:13
#>  6: 27382     FL 5S     5400  005 2020-04-27 14:57:01 2020-04-27 19:00:19
#>  7: 27396     FL 25      288  025 2020-04-27 16:10:19 2020-04-28 16:10:16
#>  8: 27397      FL 5     1080  005 2020-04-27 16:17:13 2020-04-27 17:28:04
#>  9: 27393      FL 5      864  005 2020-04-27 17:28:04 2020-04-27 18:53:48
#> 10: 27392      FL 5     5402  005 2020-04-27 18:53:48 2020-04-28 08:46:56
#> 11: 27383     FL 5S     3240  005 2020-04-27 19:00:19 2020-04-27 21:20:05
#> 12: 27384     FL 5S      864  005 2020-04-27 21:20:05 2020-04-27 21:47:18
#> 13: 27385     FL 5S     5293  005 2020-04-27 21:47:18 2020-04-28 08:35:01
#> 14: 27386     FL 5S     2700  005 2020-04-28 08:35:01 2020-04-28 10:38:11
#> 15: 27398      FL 5    11547  001 2020-04-28 08:46:56 2020-04-29 07:49:09
#> 16: 27409     FL 5S     4427  005 2020-04-28 10:38:11 2020-04-28 12:22:27
#> 17: 27410     FL 5S     3672  005 2020-04-28 12:22:27 2020-04-28 14:11:06
#> 18: 27411     FL 5S      864  005 2020-04-28 14:11:06 2020-04-28 14:54:44
#> 19: 27412     FL 5S     2651  005 2020-04-28 14:54:44 2020-04-28 16:58:00
#> 20: 27416     FL 25       96  025 2020-04-28 16:10:16 2020-04-28 17:28:01
#> 21: 27420     FL 5S      648  005 2020-04-28 16:58:00 2020-04-28 17:21:31
#> 22: 27421     FL 5S     1620  005 2020-04-28 17:21:31 2020-04-28 19:13:35
#> 23: 27417     FL 25       96  025 2020-04-28 17:28:01 2020-04-28 18:56:58
#> 24: 27418     FL 25      480  025 2020-04-28 18:56:58 2020-04-28 20:28:49
#> 25: 27432     FL 5S     2160  005 2020-04-28 19:13:35 2020-04-28 20:23:25
#> 26: 27433     FL 5S     1363  005 2020-04-28 20:23:25 2020-04-28 21:45:52
#> 27: 27419     FL 25      480  025 2020-04-28 20:28:49 2020-04-29 15:00:23
#> 28: 27413     FL 5S     4320  005 2020-04-28 21:45:52 2020-04-29 08:06:58
#> 29: 27399      FL 5    11528  001 2020-04-29 07:49:09 2020-04-29 17:16:43
#> 30: 27414     FL 5S     4320  005 2020-04-29 08:06:58 2020-04-29 10:19:20
#> 31: 27415     FL 5S      864  005 2020-04-29 10:19:20 2020-04-29 10:53:15
#> 32: 27428     FL 5S    12745  005 2020-04-29 10:53:15 2020-04-29 17:04:01
#> 33: 27424     FL 25      600  025 2020-04-29 15:00:23 2020-04-29 19:33:54
#> 34: 27429     FL 5S     3782  005 2020-04-29 17:04:01 2020-04-29 20:14:55
#> 35: 27431      FL 5     6373  005 2020-04-29 17:16:43 2020-04-30 10:15:08
#>     Order Prod_line Produced Size             Started               Ended
#>      Recipe
#>  1:        
#>  2:     005
#>  3:     005
#>  4:     005
#>  5:     005
#>  6:     005
#>  7: 005,001
#>  8:     005
#>  9:     005
#> 10:     005
#> 11:     005
#> 12:     005
#> 13:     005
#> 14: 005,001
#> 15:     001
#> 16:     001
#> 17:     001
#> 18:     001
#> 19:     001
#> 20:     001
#> 21:     001
#> 22:     001
#> 23:     001
#> 24:     001
#> 25:     001
#> 26:     001
#> 27:     001
#> 28:     001
#> 29:     001
#> 30:     001
#> 31:     001
#> 32:     001
#> 33: 001,005
#> 34: 001,005
#> 35:     005
#>      Recipe

^{由reprex package (v0.3.0) 于 2020 年 12 月 18 日创建}

【讨论】：

谢谢！这实际上给了我一个额外的功能，我没想到你可以在其中看到哪些订单与 FL 5 上的尺寸变化重叠。我相信，我希望 type = "any" 有重叠，因为实际上任何重叠都与尺寸顺序“ 001" 需要是配方 = "001"。