使用基于时间戳的时间戳在 R 中创建整齐的数据 "size"

Question

我正在分析与我们生产的不同产品相关的每个流程的周期时间变化。我们的 SAP 数据包含工人的开始和完成日志条目，objective 用于使用此信息计算周期时间。

但是，SAP 在一列中导出开始和结束时间戳，并且没有可用的参考列说明开始时间和结束时间。这使得用例如整理数据无法传播。

当前数据

160 万行
150 次操作
1万个订单

。一小部分数据样本如下所示。

Order <-  rep(c(1059866,1059891),each = 4)
Operation <- rep(c(1510,1550),4)
Timestamp <- c("30-11-2016 07:33:30", "30-11-2016 07:33:42", "30-11-2016 16:00:13", "30-11-2016 16:00:18", "30-11-2016 07:35:21", "30-11-2016 07:35:43", "30-11-2016 16:00:43", "30-11-2016 16:00:39")

df_current <- cbind(Order, Operation, Timestamp)

每个流程步骤都需要此开始和结束信息 ("Operation")。从逻辑上讲，最早的时间戳是开始日志条目，最晚的时间戳是完成日志条目。

但是我不知道如何告诉 R 创建一个新列，根据时间戳正确指示哪个时间戳开始和结束。

所需数据

Order <-  rep(c(1059866,1059891),each = 4)
Operation <- rep(c(1510,1550),4)
Timestamp <- c("30-11-2016 07:33:30", "30-11-2016 07:33:42", "30-11-2016 16:00:13", "30-11-2016 16:00:18", "30-11-2016 07:35:21", "30-11-2016 07:35:43", "30-11-2016 16:00:43", "30-11-2016 16:00:39")
Status <- c("Start" , "Finish", "Start" , "Finish", "Start" , "Finish",  "Finish", "Start")   

df_desired <- cbind(Order, Operation, Timestamp, Status)

当数据看起来像那样时，我可以轻松整理数据。

谢谢

Answer 1

假设您可以将数据转换为 data.frame 而不是 matrix:

df_current <- data.frame(Order, Operation, Timestamp)

df.With.Status <- do.call(rbind, #rbind the dataframes to a big dataframe
lapply(split(df_current,list(df_current$Order,df_current$Operation)), #split dataframe by unique order/operation combinations and apply function for each combination
       function(df){
         df$Timestamp <- strptime(rev(as.character(df$Timestamp)),format="%d-%m-%Y %H:%M:%S") #Convert to Time, so that it is sortable
         df <- df[order(df$Timestamp),] # rearrange the dataframe in case of wrong order
         df$Status <- c("Start","Finish") #add status
         return(df)
       }))

Answer 2

和dplyr


library(dplyr)


  df_current %>% as.data.frame() %>%
    group_by(Operation, Order) %>%
    mutate(Timestamp = as.POSIXct(Timestamp, format = "%d-%m-%Y %H:%M:%S"),
           Status = case_when(Timestamp == min(Timestamp) ~ "Start",
                              TRUE ~ "Finish")) %>%
    arrange(Order, Operation)


# A tibble: 8 x 4
# Groups:   Operation, Order [4]
  Order   Operation Timestamp           Status
  <fct>   <fct>     <dttm>              <chr> 
1 1059866 1510      2016-11-30 07:33:30 Start 
2 1059866 1510      2016-11-30 16:00:13 Finish
3 1059866 1550      2016-11-30 07:33:42 Start 
4 1059866 1550      2016-11-30 16:00:18 Finish
5 1059891 1510      2016-11-30 07:35:21 Start 
6 1059891 1510      2016-11-30 16:00:43 Finish
7 1059891 1550      2016-11-30 07:35:43 Start 
8 1059891 1550      2016-11-30 16:00:39 Finish

此外，由于您的数据很大：data.table

library(data.table)

dfc_2 <- as.data.frame(df_current)

dfc_2$Timestamp <- as.POSIXct(Timestamp, format = "%d-%m-%Y %H:%M:%S")

setDT(df_curr)[, Status := ifelse(Timestamp == min(Timestamp), "Start", "Finish"),
               keyby = .(Operation, Order)]

使用基于时间戳的时间戳在 R 中创建整齐的数据 "size"

Create tidy data in R with time stamps based on time stap "size"

sap

r