R:如何根据多列条件聚合(和汇总)df 中的行并保持先前​​的顺序？

Question

我想在以下数据示例中聚合行(并汇总值):

    df <- data.frame(from=c("A" ,"A", "A", "C", "C", "D", "A"),
                 to=c("B", "B", "B", "A", "A", "B", "D"),
                 values=c(5,6,2,10,2,6,3),
                 product=c("x","x", "x", "y", "z", "w", "w"),
                 year=c(1990,1991,1991,1990,1990,1991,1992))

> df
  from to values product year
1    A  B      5       x 1990
2    A  B      6       x 1991
3    A  B      2       x 1991
4    C  A     10       y 1990
5    C  A      2       z 1990
6    D  B      6       w 1991
7    A  D      3       w 1992

from、to、product 和 year 列的所有包含相同值/字符的行应该是聚合到一行，values 列中的值应该相加。

我尝试了以下代码:

aggregate(values~from+to+product+year, df, FUN=sum)

和

ddply(df_id, c("from", "to", "product", "year"), numcolwise(sum))

这些代码运行良好。但是，两者都更改了行的顺序(还有不太重要的列)，请参见下文:

for aggregate:
      from to product year values
    1    A  B       x 1990      5
    2    C  A       y 1990     10
    3    C  A       z 1990      2
    4    D  B       w 1991      6
    5    A  B       x 1991      8
    6    A  D       w 1992      3

and for ddply:

       from    to product year values
    1     C     A       y 1990     10
    2     C     A       z 1990      2
    3     A     B       x 1990      5
    4     A     B       x 1991      8
    5     A     D       w 1992      3
    6     D     B       w 1991      6

预期的结果应该是这样的:

  from to values product year
1    A  B      5       x 1990
2    A  B      8       x 1991
3    C  A     10       y 1990
4    C  A      2       z 1990
5    D  B      6       w 1991
6    A  D      3       w 1992

关于如何解决这个顺序问题(至少对于行)有什么想法吗？谢谢

Answer 1

data.table 包默认保留原始分组排序:

library(data.table)

setDT(df)[, .(v = sum(values)), by=.(from,to,product,year)] 

#    from to product year  v
# 1:    A  B       x 1990  5
# 2:    A  B       x 1991  8
# 3:    C  A       y 1990 10
# 4:    C  A       z 1990  2
# 5:    D  B       w 1991  6
# 6:    A  D       w 1992  3

只有 keyby= 而不是 by= 才会对组进行排序。