如何在计算每个用户的第一个和最后一个日期之间的差异时修复错误？

Question

我想计算用户旅程的第一个接触点和用户旅程的最后一个接触点之间的差异，以及所有旅程的接触点之间的差异。

这是CJ数据的（简短）示例：

PurchaseID   timestamp                date
1            2016-03-12 22:18:34      2016-03-12
1            2016-03-13 05:25:49      2016-03-13
2            2015-07-18 13:00:38      2015-07-18
2            2015-08-07 19:16:59      2015-08-07
2            2015-11-03 12:31:35      2015-11-03
...

我想创建一个新的变量difference，它是每个购买ID的第一个和最后一个日期之间的差异。

根据本网站上的其他文章，我尝试过并且应该工作的内容如下：

# difference 
CJ <- data.table(CJ)
CJ[, difference := max(timestamp) - min(timestamp), by = PurchaseID]

这给出了一个错误：

Error in `[.data.frame`(CJ, , `:=`(diff, max(timestamp) - min(timestamp)),  : 
  unused argument (by = PurchaseID)

当我只使用变量date时会发生同样的错误。在我的数据子集中，未发生此错误。到目前为止，我找不到根本原因。有什么想法吗？

另外，dput的输出

> dput(head(CJgroup))
structure(list(UserID = c(9558L, 9558L, 9558L, 9657L, 1L, 1L), 
    PurchaseID = c(1L, 1L, 1L, 2L, 3L, 4L), timestamp = structure(c(1457817514, 
    1457843149, 1457868381, 1437217238, 1438967819, 1446550295
    ), class = c("POSIXct", "POSIXt"), tzone = "Europe/Amsterdam"), 
    duration = c(5.786, 65.725, 6.492, 57, 120, 459), device = structure(c(2L, 
    2L, 2L, 1L, 1L, 1L), .Label = c("FIXED", "MOBILE"), class = "factor"), 
    touchpoint = c(7L, 7L, 7L, 4L, 7L, 1L), purchase_own = c(0L, 
    0L, 0L, 0L, 0L, 0L), purchase_any = c(0L, 0L, 0L, 0L, 0L, 
    0L), MobilePanel = c(0L, 0L, 0L, 0L, 0L, 0L), FixedPanel = c(0L, 
    0L, 0L, 0L, 17L, 17L), CIT = c(0, 0, 0, 0, 0, 0), FIT = c(1, 
    1, 1, 1, 1, 1), T1 = c(0, 0, 0, 0, 0, 1), T2 = c(0, 0, 0, 
    0, 0, 0), T3 = c(0, 0, 0, 0, 0, 0), T4 = c(0, 0, 0, 1, 0, 
    0), T5 = c(0, 0, 0, 0, 0, 0), T6 = c(0, 0, 0, 0, 0, 0), T7 = c(1, 
    1, 1, 0, 1, 0), T8 = c(0, 0, 0, 0, 0, 0), T9 = c(0, 0, 0, 
    0, 0, 0), T10 = c(0, 0, 0, 0, 0, 0), T12 = c(0, 0, 0, 0, 
    0, 0), T13 = c(0, 0, 0, 0, 0, 0), T14 = c(0, 0, 0, 0, 0, 
    0), T15 = c(0, 0, 0, 0, 0, 0), T16 = c(0, 0, 0, 0, 0, 0), 
    T18 = c(0, 0, 0, 0, 0, 0), T19 = c(0, 0, 0, 0, 0, 0), T20 = c(0, 
    0, 0, 0, 0, 0), T21 = c(0, 0, 0, 0, 0, 0), T22 = c(0, 0, 
    0, 0, 0, 0), devicemobile = c(1, 1, 1, 0, 0, 0), devicefixed = c(0, 
    0, 0, 1, 1, 1), purchase_comp = c(0, 0, 0, 0, 0, 0), date = structure(c(16872, 
    16873, 16873, 16634, 16654, 16742), class = "Date"), POS_comp = c(0, 
    0, 0, 0, 0, 0), POS_own = c(0, 0, 0, 0, 0, 0), CountTP = c(1L, 
    2L, 3L, 1L, 1L, 1L)), row.names = c(NA, 6L), class = "data.frame")

Answer 1

这是一个使用dplyr包而不是data.table的解决方案。

然后你可以做以下事情

library(dplyr)


CJgroup %>% select(PurchaseID, date) %>% 
  group_by(PurchaseID) %>% 
  summarise(difference = as.numeric(max(date) - min(date)))

# A tibble: 4 x 2
  PurchaseID difference
       <int>      <dbl>
1          1          1
2          2          0
3          3          0
4          4          0