Question

我有一个数据集df

 Read      Box       ID      Time                             Subject 
 T         out               10/1/2019 9:00:01 AM
 T         out               10/1/2019 9:00:02 AM             Re:
 T         out               10/1/2019 9:00:03 AM             Re:
 T         out               10/1/2019 9:02:59 AM             Re:
 T         out               10/1/2019 9:03:00 AM
 F                           10/1/2019 9:05:00 AM
 T         out               10/1/2019 9:06:00 AM             Fwd:
 T         out               10/1/2019 9:06:02 AM             Fwd:
 T         in                10/1/2019 9:07:00 AM
 T         in                10/1/2019 9:07:02 AM
 T         out               10/1/2019 9:07:04 AM
 T         out               10/1/2019 9:07:05 AM             Fw:
 T         out               10/1/2019 9:07:06 AM             Fw:
           hello             10/1/2019 9:07:08 AM

根据此数据集中的某些条件，我想创建一个startime列和一个endtime列。

在发生以下情况时，我想创建一个“开始时间”：如果“主题”列的第一个单词以RE：，re，FWD或FW（以连续方式）开头，则Read ==“ T”，Box = =“ out”和ID ==“”

第一次出现这种情况时，将生成一个开始时间。例如，对于此数据集，开始时间将为10/1/2019 9:00:02 AM，因为这是我们首先看到所需条件的位置（主题为FW：，RE：或FWD，Read = T，Box =出，ID =“”“）但是，当这些条件中的任何一个都不成立时，将创建结束时间。因此，第一个结束时间将发生在第4行之前，该时间为10/1/2019 9:02:59 AM。我的最终目标是为此创建一个工期列。

当包含RE，Fwd和Fw时，这是我想要的输出

  starttime                    endtime                     duration

  10/1/2019 9:00:02 AM        10/1/2019 9:02:59 AM         177 secs
  10/1/2019 9:06:00 AM        10/1/2019 9:06:02 AM         2 secs
  10/1/2019 9:07:05 AM        10/1/2019 9:07:06 AM         1 secs

此外，我将如何在单独的代码中指定这些条件的开始时间和结束时间：读取= T，方框=输出，ID =“”，并且主题列的第一个单词不包含Re，Fwd或Fw？

 Read      Box       ID      Time                             Subject 
 T         out               10/1/2019 9:00:01 AM
 T         out               10/1/2019 9:00:02 AM             Re:
 T         out               10/1/2019 9:00:03 AM             Re:
 T         out               10/1/2019 9:02:59 AM             Re:
 T         out               10/1/2019 9:03:00 AM
 F                           10/1/2019 9:05:00 AM
 T         out               10/1/2019 9:06:00 AM             Fwd:
 T         out               10/1/2019 9:06:02 AM             Fwd:
 T         in                10/1/2019 9:07:00 AM
 T         in                10/1/2019 9:07:02 AM
 T         out               10/1/2019 9:07:04 AM
 T         out               10/1/2019 9:07:05 AM             Fw:
 T         out               10/1/2019 9:07:06 AM             Fw:
           hello             10/1/2019 9:07:08 AM

这是排除RE，Fwd和Fw时我想要的输出

  starttime                    endtime                     duration

  10/1/2019 9:00:01 AM        10/1/2019 9:00:01 AM         0 secs
  10/1/2019 9:03:00 AM        10/1/2019 9:03:00 AM         0 secs
  10/1/2019 9:07:04 AM        10/1/2019 9:07:04 AM         0 secs

dput：

 structure(list(Read = structure(c(3L, 3L, 3L, 3L, 3L, 2L, 3L, 
3L, 3L, 3L, 4L, 4L, 3L, 1L), .Label = c("", "F", "T", "T "), class = "factor"), 
Box = structure(c(3L, 3L, 3L, 3L, 3L, 1L, 3L, 3L, 2L, 2L, 
3L, 3L, 3L, 1L), .Label = c("", "in", "out"), class = "factor"), 
ID = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 2L), .Label = c("", "hello"), class = "factor"), 
Time = structure(1:14, .Label = c("10/1/2019 9:00:01 AM", 
"10/1/2019 9:00:02 AM", "10/1/2019 9:00:03 AM", "10/1/2019 9:02:59 AM", 
"10/1/2019 9:03:00 AM", "10/1/2019 9:05:00 AM", "10/1/2019 9:06:00 AM", 
"10/1/2019 9:06:02 AM", "10/1/2019 9:07:00 AM", "10/1/2019 9:07:02 AM", 
"10/1/2019 9:07:04 AM", "10/1/2019 9:07:05 AM", "10/1/2019 9:07:06 AM", 
"10/1/2019 9:07:08 AM"), class = "factor"), Subject = structure(c(1L, 
4L, 4L, 4L, 1L, 1L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 1L), .Label = c("", 
"Fw:", "Fwd:", "Re:"), class = "factor")), class = "data.frame", row.names = c(NA, 
-14L))

建议的代码有效，我也想同时包含Subject列条件：
其中Subject == FW，FWD，RE（忽略大写/小写）和如果Subject不等于FW，FWD，Re（忽略大小写）

library(dplyr)

df %>%
mutate(Time = lubridate::mdy_hms(Time), 
cond = Read == "T" & Box == "out" & ID == "" & Subject == "FW" & Subject  == "FWD" & Subject == "RE" (ignore.case = TRUE)
grp = cumsum(!cond)) %>%
filter(cond) %>%
group_by(grp) %>%
summarise(starttime = first(Time), 
endtime = last(Time), 
duration = difftime(endtime, starttime, units = "secs")) %>%
select(-grp)

库（dplyr）

df %>%
mutate(Time = lubridate::mdy_hms(Time), 
cond = Read == "T" & Box == "out" & ID == "" & Subject! == "FW" & Subject! == "FWD" & Subject! == "RE" (ignore.case = TRUE)
grp = cumsum(!cond)) %>%
filter(cond) %>%
group_by(grp) %>%
summarise(starttime = first(Time), 
endtime = last(Time), 
duration = difftime(endtime, starttime, units = "secs")) %>%
select(-grp)

Answer 1

您的问题的整个部分已经在您的其他问题（Create start and endtime columns based on multiple conditions in R (dplyr, lubridate)）中得到了回答。我知道这很困难，但是下次请着重于您尚不了解的问题，将问题缩小到较小的范围。

如果要检测子字符串，最好的方法是使用str_detect包（stringr的一部分）中的tidyverse：

library(tidyverse)
library(lubridate)
df %>%
  mutate(Time = mdy_hms(Time), 
         # cond = Read == "T" & Box == "out" & ID == "", #from the answer https://stackoverflow.com/a/60068929/3888000
         cond = Read == "T" & Box == "out" & ID == "" & str_detect(Subject, regex('FW|FWD|RE', ignore_case=TRUE)), #including those subjects
         # cond = Read == "T" & Box == "out" & ID == "" & !str_detect(Subject, regex('FW|FWD|RE', ignore_case=TRUE)), #excluding those subjects
         grp = cumsum(!cond)) %>%
  filter(cond) %>%
  group_by(grp) %>%
  summarise(starttime = first(Time), 
            endtime = last(Time), 
            duration = difftime(endtime, starttime, units = "secs")) %>%
  select(-grp)

这使用正则表达式（regex），这是一件非常好的事情。该代码只有OR（|）运算符，非常易于阅读，但可能性是无限的。

根据多种过滤条件（R，dplyr）创建时间戳列

1 个答案: