如何在data.table中链序、分组、变异、支点？

Question

我是 data.table 的新手&我试图复制一些dplyr代码，但在data.table代码中存在支点/重塑数据的问题。

libs

library(data.table)
library(lubridate)
library(tidyverse)

df

test_df <- data.frame(id = c(1234, 1234, 5678, 5678),
           date = c("2021-10-10","2021-10-10", "2021-8-10", "2021-8-15"),
           Amount  = c(54767, 96896, 34534, 79870)) %>% 
  
  mutate(date = ymd(date))

dplyr代码：

test_df %>% 
  group_by(id) %>% 
  arrange(date) %>% 
  mutate(Amt_first = first(Amount),
         Amt_last = last(Amount)) %>%
  ungroup() %>%

  pivot_longer(cols = c(Amt_first:Amt_last), names_to = "Amt_catg", 
  values_to = "Amt_val")

结果：

# A tibble: 8 x 5
     id date       Amount Amt_catg  Amt_val
  <dbl> <date>      <dbl> <chr>       <dbl>
1  5678 2021-08-10  34534 Amt_first   34534
2  5678 2021-08-10  34534 Amt_last    79870
3  5678 2021-08-15  79870 Amt_first   34534
4  5678 2021-08-15  79870 Amt_last    79870
5  1234 2021-10-10  54767 Amt_first   54767
6  1234 2021-10-10  54767 Amt_last    96896
7  1234 2021-10-10  96896 Amt_first   54767
8  1234 2021-10-10  96896 Amt_last    96896

data.table尝试

setDT(test_df)[order(date),
             `:=`(Amt_first = data.table::first(Amount),
                   Amt_last = data.table::last(Amount)), 
             by = id] %>% 
    
    # https://cran.r-project.org/web/packages/data.table/vignettes/datatable-reshape.html
  data.table::melt(measure.vars = c("Amt_first","Amt_Last"),
                   variable.name = "Amt_catg", value.name = "Amt_val")

我这里有个错误。

Answer 1

您会得到一个错误，因为您拼写错了Amt_last：

setDT(test_df)[order(date),
                 `:=`(Amt_first = data.table::first(Amount),
                      Amt_last = data.table::last(Amount)), 
                 by = id] %>%
  data.table::melt(id.vars=c('id','date'),measure.vars = c("Amt_first","Amt_last"),
                   variable.name = "Amt_catg", value.name = "Amt_val")

     id       date Amount  Amt_catg Amt_val
1: 1234 2021-10-10  54767 Amt_first   54767
2: 1234 2021-10-10  96896 Amt_first   54767
3: 5678 2021-08-10  34534 Amt_first   34534
4: 5678 2021-08-15  79870 Amt_first   34534
5: 1234 2021-10-10  54767  Amt_last   96896
6: 1234 2021-10-10  96896  Amt_last   96896
7: 5678 2021-08-10  34534  Amt_last   79870
8: 5678 2021-08-15  79870  Amt_last   79870

Answer 2

输出可以通过完全连接来实现：

setDT(test_df)

merge(
  test_df,
  test_df[order(date), 
          .(Amt_catg = c('Amt_first', 'Amt_last'), 
            Amt_val = Amount[c(1L, .N)]), 
          by = id]
)

#      id       date Amount  Amt_catg Amt_val
# 1: 1234 2021-10-10  54767 Amt_first   54767
# 2: 1234 2021-10-10  54767  Amt_last   96896
# 3: 1234 2021-10-10  96896 Amt_first   54767
# 4: 1234 2021-10-10  96896  Amt_last   96896
# 5: 5678 2021-08-10  34534 Amt_first   34534
# 6: 5678 2021-08-10  34534  Amt_last   79870
# 7: 5678 2021-08-15  79870 Amt_first   34534
# 8: 5678 2021-08-15  79870  Amt_last   79870