在R中重塑活动日志

我有一个存储在数据框中的活动日志，记录了代理接听电话的时间：

Agent   | Call Start |  Call End
--------------------------------
Albert  | 9:33:31    | 9:38:30

Albert  | 9:40:33    | 9:44:36

Boris   | 9:27:39    | 9:36:39

等等

我想重新调整数据，以便我可以看到在一天中的任何时间段内哪些代理被占用或可用：

From    |   To      |   Albert |    Boris
-----------------------------------
9:00:01 | 9:27:38   | Free     | Free

9:27:39 | 9:33:30   | Free     | Occupied

9:33:31 | 9:36:39   | Occupied | Occupied

9:36:39 | 9:38:30   | Occupied | Free

等等

---

如果可能的话，我希望使用

tidyverse

工具来实现这一点，但对任何有用的工具都开放

恐怕我对

tidyverse

不是很了解，但这应该会产生你想要的结果

x    
 Agent Call_Start Call_End
1 Albert    9:33:31  9:38:30
2 Albert    9:40:33  9:44:36
3  Boris    9:27:39  9:36:39

x$Call_Start<-as.POSIXct(x$Call_Start, format=("%H:%M:%S"))
x$Call_End<-as.POSIXct(x$Call_End, format=("%H:%M:%S"))

#create a list of times to check 
start<-as.POSIXct("09:27:00", format=("%H:%M:%S"))
end<-as.POSIXct("09:45:00", format=("%H:%M:%S"))

t<-seq.POSIXt(start, end, by="sec")

#create a dataframe to hold the results
agents<-unique(x$Agent)

occ_dat<-data.frame(matrix(nrow=length(t), ncol=length(agents)+1))

occ_dat[,1]<-t

names(occ_dat)<-c("time", as.character(agents))

#divide the data up by agent
each_agent<-split(x, f=x$Agent)

#check for each agent whether or not they were occupied at each time
sil<-sapply(seq_along(agents), function(i){

  c_agent<-each_agent[[i]]

  apply(c_agent, 1, function(call){

    occ_dat[,(i+1)]<<-ifelse(occ_dat$time>=call[2] & 
                         occ_dat$time<=call[3], "occupied", occ_dat[,(i+1)])
  })
})

#if agent is not occupied set them to "free"
occ_dat[is.na(occ_dat)]<-"free"

#collapse all times into non-overlapping time periods by giving a unique ID
#every time either agent has a status change

occ_dat$status<-1

counter<-1

occ_dat$status[2:nrow(occ_dat)]<-sapply(2:nrow(occ_dat), function(i){
  if(occ_dat[i,2]==occ_dat[i-1,2] &
     occ_dat[i,3]==occ_dat[i-1,3]) {counter}
  else {counter<<-counter+1; counter}
})

#use duplicated to get the start and end of each unique status period and combine
od1<-occ_dat[!duplicated(occ_dat$status),]

od2<-occ_dat[!duplicated(occ_dat$status, fromLast = TRUE),]

od3<-merge(od1, od2[,c("time", "status")], by="status")

#some formatting
od3<-od3[,c(2,5,3,4)]
names(od3)[1:2]<-c("From", "To")

od3
               From                  To   Albert    Boris
1 2018-06-05 09:27:00 2018-06-05 09:27:38     free     free
2 2018-06-05 09:27:39 2018-06-05 09:33:30     free occupied
3 2018-06-05 09:33:31 2018-06-05 09:36:39 occupied occupied
4 2018-06-05 09:36:40 2018-06-05 09:38:30 occupied     free
5 2018-06-05 09:38:31 2018-06-05 09:40:32     free     free
6 2018-06-05 09:40:33 2018-06-05 09:44:36 occupied     free
7 2018-06-05 09:44:37 2018-06-05 09:45:00     free     free

x
代理呼叫\u开始呼叫\u结束
1阿尔伯特9:33:31 9:38:30
2阿尔伯特9:40:33 9:44:36
3鲍里斯9:27:39 9:36:39
x$Call_Start好的-这里的答案充分利用了tidyverse的软件包和功能，尤其是spread和gather功能

library(tidyr)
library(dplyr)
library(chron)
library(zoo)

#Create sample activity log
Agent <- c('Albert','Albert','Boris','Boris','Charlie','Charlie','Charlie')
Call_Start <- times(c('9:33:31','9:40:33','9:27:39','9:41:23','9:19:09','9:35:28','9:46:31')) #use chron function
Call_End <- times(c('9:38:30','9:44:36','9:36:39','9:42:29','9:19:19','9:45:28','9:46:55'))
Start <- data.frame(Agent,Call_Start,Call_End)

#Set the Start and End times for the Day 
start_time <- times(c('09:00:01'))
end_time <- times(c('17:00:01'))

#Create the specific time breaks needed
Time_breaks_start <- sort(c(start_time,Start$Call_Start,Start$Call_End))
Time_breaks_end <- sort(c(Start$Call_Start, Start$Call_End,end_time))
Time_breaks <- data.frame(Time_breaks_start,Time_breaks_end)

#Use gather and spread to identify when agents start and end calls.
Gathered <- Start %>% 
  gather(key="Start_End",value="Time_breaks_start", Call_Start, Call_End) %>% 
  mutate(Start_End = ifelse(Start_End == 'Call_Start','Occupied','Free'))
Spread <- Gathered %>% 
        spread(key=Agent, value=Start_End)

#Join spread data to list of time breaks and clean up missing values
Output <- Time_breaks %>% 
      left_join(Spread, by='Time_breaks_start') %>%
      do(na.locf(.)) %>% #uses function from zoo
      apply(2, function(x) { x[is.na(x)] <- "Free" ; x })

library（tidyr）
图书馆（dplyr）
图书馆（计时）
图书馆（动物园）
#创建示例活动日志
代理你从哪里得到时间的
9:30:01
？为什么在
9:27:38
之前？这是一个打字错误。我将对其进行编辑。我的思路如下：为每个事务记录创建一个从呼叫开始到呼叫结束的时间向量（以秒为单位），将每个代理的向量组合成一个向量，将该向量与一天开始时间和结束时间之间的秒向量相结合。如果代理的向量与所有秒的向量匹配，则代理被设置为已占用；如果没有，则标记为免费。理论上听起来不错，但我关心的是（a）性能和（b）如何跨多个代理和多个日期执行。