如何使用R data.table将值从一行和一列（“单元格”）复制（“结转”）到不同的行和列？

Question

我有一个R data.table 。 我想根据一个key将值从一列和一行复制或转发到不同的列和行。 我用merge和rbindlist完成了这个，但我想知道是否有一个更简约的解决方案。

这是一个玩具示例：假设我在不同日期跟踪不同支票账户的期初和期末余额。 我希望将前一天的（ t-1 ）期末余额结转到当天的（ t ）期初余额。

有这个输入：

> DT_in <- data.table(date = c("2017-12-29", "2017-12-29", "2017-12-29", "2018-01-02", 
+                         "2018-01-02", "2018-01-02", "2018-01-02", "2018-01-03", 
+                         "2018-01-03", "2018-01-03", "2018-01-03"), 
+                  account_id = c("A17", "A23", "B21", "A17", "A23", "B21", "C12", 
+                                 "A17", "A23", "B21", "C12"),
+                  opening_balance = NA,
+                  ending_balance = c(224, 254, 240, 290, 107, 272, 105, 256, 215, 
+                                     202, 238)
+                  )
> DT_in
          date account_id opening_balance ending_balance
 1: 2017-12-29        A17              NA            224
 2: 2017-12-29        A23              NA            254
 3: 2017-12-29        B21              NA            240
 4: 2018-01-02        A17              NA            290
 5: 2018-01-02        A23              NA            107
 6: 2018-01-02        B21              NA            272
 7: 2018-01-02        C12              NA            105
 8: 2018-01-03        A17              NA            256
 9: 2018-01-03        A23              NA            215
10: 2018-01-03        B21              NA            202
11: 2018-01-03        C12              NA            238

想要这个输出：

> DT_out <- data.table(date = c("2017-12-29", "2017-12-29", "2017-12-29", "2018-01-02", 
+                           "2018-01-02", "2018-01-02", "2018-01-02", "2018-01-03", 
+                           "2018-01-03", "2018-01-03", "2018-01-03"), 
+                  account_id = c("A17", "A23", "B21", "A17", "A23", "B21", "C12", 
+                                 "A17", "A23", "B21", "C12"),
+                  opening_balance = c(NA, NA, NA, 224, 254, 240, NA, 290, 107, 272, 105), 
+                  ending_balance = c(224, 254, 240, 290, 107, 272, 105, 256, 215, 
+                                     202, 238)
+                  )
> DT_out
          date account_id opening_balance ending_balance
 1: 2017-12-29        A17              NA            224
 2: 2017-12-29        A23              NA            254
 3: 2017-12-29        B21              NA            240
 4: 2018-01-02        A17             224            290
 5: 2018-01-02        A23             254            107
 6: 2018-01-02        B21             240            272
 7: 2018-01-02        C12              NA            105
 8: 2018-01-03        A17             290            256
 9: 2018-01-03        A23             107            215
10: 2018-01-03        B21             272            202
11: 2018-01-03        C12             105            238

请注意， account_id不一定是从一个日期到下一个日期的持久性（可以添加新的或旧的删除）。 虽然没关系，但请注意，日期是营业日期而不是日历日期。

Answer 1

如果修复示例数据，则使用相同的数据类型“opening_balance”和“ending_balance”列

...
opening_balance = NA_real_,
...

您可以使用此代码向每个组执行最后一次观察（在应用正确的顺序/排序后！）：

setorder(DT_in, account_id, date)
DT_in[, opening_balance := shift(ending_balance, 1), by = .(account_id)]

这导致（排序输出）：

> DT_in
          date account_id opening_balance ending_balance
 1: 2017-12-29        A17              NA            224
 2: 2018-01-02        A17             224            290
 3: 2018-01-03        A17             290            256
 4: 2017-12-29        A23              NA            254
 5: 2018-01-02        A23             254            107
 6: 2018-01-03        A23             107            215
 7: 2017-12-29        B21              NA            240
 8: 2018-01-02        B21             240            272
 9: 2018-01-03        B21             272            202
10: 2018-01-02        C12              NA            105
11: 2018-01-03        C12             105            238