如何计算R中变量行之间的时间差?

时间:2017-03-10 01:01:36

标签: r data.table difftime

我希望根据开始工作时间和结束工作时间计算不同组的时间差异。如何根据组中的标签告诉R计算两行之间的difftime?以下是一个示例数据集:

library(data.table)


latemail <- function(N, st="2012/01/01", et="2012/02/01") {
  st <- as.POSIXct(as.Date(st))
  et <- as.POSIXct(as.Date(et))
  dt <- as.numeric(difftime(et,st,unit="sec"))
  ev <- sort(runif(N, 0, dt))
  rt <- st + ev

}

#create our data frame
set.seed(42)
dt = latemail(20)
work = setDT(as.data.frame(dt))
work[,worker:= stringi::stri_rand_strings(2, 5)]  
work[,dt:= as.POSIXct(as.character(work$dt), tz = "GMT")]
work[,status:=NA]

#order
setorder(work, worker, dt)

#add work times
work$status[1] = "start"
work$status[5] = "end"
work$status[6] = "start"
work$status[10] = "end"
work$status[11] = "start"
work$status[15] = "end"
work$status[16] = "start"
work$status[20] = "end"

表现在看起来像这样:

                    dt worker status
 1: 2012-01-04 23:11:31  VOuRp  start
 2: 2012-01-09 15:53:16  VOuRp     NA
 3: 2012-01-15 02:56:45  VOuRp     NA
 4: 2012-01-16 21:12:26  VOuRp     NA
 5: 2012-01-20 16:27:31  VOuRp    end
 6: 2012-01-22 15:34:05  VOuRp  start
 7: 2012-01-23 15:01:18  VOuRp     NA
 8: 2012-01-29 03:36:56  VOuRp     NA
 9: 2012-01-29 20:11:02  VOuRp     NA
10: 2012-01-31 02:48:01  VOuRp    end
11: 2012-01-04 10:24:38  u8zw5  start
12: 2012-01-08 17:02:20  u8zw5     NA
13: 2012-01-14 23:33:35  u8zw5     NA
14: 2012-01-15 12:23:52  u8zw5     NA
15: 2012-01-18 03:53:15  u8zw5    end
16: 2012-01-21 03:48:08  u8zw5  start
17: 2012-01-23 02:01:10  u8zw5     NA
18: 2012-01-26 12:51:10  u8zw5     NA
19: 2012-01-29 18:23:46  u8zw5     NA
20: 2012-01-29 22:22:14  u8zw5    end

我正在寻找答案: 最终我想获得最低值(标记为worker 1和worker 2只是因为不知道如何为stringi执行set.seed()的并行操作)。下面的代码给出了worker 1的第一行,但是我希望每个worker都有一个班次:

difftime(as.POSIXct("2012-01-20 16:27:31"), as.POSIXct("2012-01-04 23:11:31"), units = "hours")
    Work time   time difference in hours  
    worker 1         377.2667 hours
    worker 2         . . . . 

在这个例子中,我在worker之间有一组偶数值,但假设我在不同worker之间有可变行,那会是什么样子?我假设某种difftime公式?当我处理大数据时,我会使用数据表解决方案。

1 个答案:

答案 0 :(得分:4)

以下是使用data.table的解决方案:

 work[status %in% c("start", "end"), 
        time.diff := ifelse(status == "start", 
        difftime(shift(dt, fill = NA, type = "lead"), dt, units = "hours"), NA), 
        by = worker][status == "start", sum(time.diff), worker]

我们得到:

 worker       V1
1:  VOuRp 580.4989
2:  u8zw5 540.0453
> 

其中V1具有每个工人的起始间隔的所有小时数之和。

让我们一步一步地解释它,以便更好地理解。

第1步。选择startend状态的所有行:

work.se <- work[status %in% c("start", "end")]

                    dt worker status
1: 2012-01-04 23:11:31  VOuRp  start
2: 2012-01-20 16:27:31  VOuRp    end
3: 2012-01-22 15:34:05  VOuRp  start
4: 2012-01-31 02:48:01  VOuRp    end
5: 2012-01-04 10:24:38  u8zw5  start
6: 2012-01-18 03:53:15  u8zw5    end
7: 2012-01-21 03:48:08  u8zw5  start
8: 2012-01-29 22:22:14  u8zw5    end
> 

第2步:创建一个函数,用于计算当前行与下一行之间的时差。将在data.table对象内调用此函数。我们使用同一个包中的shift函数:

getDiff <- function(x) {
    difftime(shift(x, fill = NA, type = "lead"), x, units = "hours")
}

getDiff计算与下一条记录(组内)和当前记录之间的时差。它为最后一行指定NA,因为没有下一个值。然后我们从计算中排除NA值。

第3步:在data.table语法中调用它:

work.result <- work.se[, time.diff := ifelse(status == "start", 
    getDiff(dt), NA), by = worker]

我们得到了这个:

                    dt worker status time.diff
1: 2012-01-04 23:11:31  VOuRp  start  377.2667
2: 2012-01-20 16:27:31  VOuRp    end        NA
3: 2012-01-22 15:34:05  VOuRp  start  203.2322
4: 2012-01-31 02:48:01  VOuRp    end        NA
5: 2012-01-04 10:24:38  u8zw5  start  329.4769
6: 2012-01-18 03:53:15  u8zw5    end        NA
7: 2012-01-21 03:48:08  u8zw5  start  210.5683
8: 2012-01-29 22:22:14  u8zw5    end        NA

第4步:为每个工作人员总结NA列的非time.diff值:

> work.result[status == "start", sum(time.diff), worker]
   worker       V1
1:  VOuRp 580.4989
2:  u8zw5 540.0453
> 

data.table对象可以通过附加的[]进行连接,因此可以合并为最后一部分的单个句子:

work.se[, time.diff := ifelse(status == "start", 
    getDiff(dt), NA), by = worker][status == "start", sum(time.diff), worker]

最终:将所有内容合并为一句话:

work[status %in% c("start", "end"), 
    time.diff := ifelse(status == "start", 
    difftime(shift(dt, fill = NA, type = "lead"), dt, units = "hours"), NA), 
    by = worker][status == "start", sum(time.diff), worker]

检查此link以获取data.table基本语法。 我希望这会有所帮助,如果这是你想要的,请告诉我们