在data.table

Question

總和選擇性行我有一個數據幀如下

structure(list(code = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), Month = structure(c(4L, 
3L, 7L, 1L, 8L, 6L, 5L, 2L, 9L, 4L, 3L, 7L, 1L, 8L, 6L, 5L, 2L, 
9L), .Label = c("Apr", "Aug", "Feb", "Jan", "Jul", "Jun", "Mar", 
"May", "Sep"), class = "factor"), Var1 = c(10L, 20L, 30L, 40L, 
50L, 60L, 70L, 80L, 90L, 10L, 20L, 30L, 40L, 50L, 60L, 70L, 80L, 
90L), var2 = c(2, 3.5, 55, 23.5, 1, 1, 1, 1.5, 1, 1.5, 1, 1, 
1, 1.5, 1.75, 1.75, 1, 1)), .Names = c("code", "Month", 
"Var1", "var2"), class = "data.frame", row.names = c(NA, -18L 
)) 

      code Month Var1 var2 
1    0 Jan 10 2.00 
2    0 Feb 20 3.50 
3    0 Mar 30 55.00 
4    0 Apr 40 23.50 
5    0 May 50 1.00 
6    0 Jun 60 1.00 
7    0 Jul 70 1.00 
8    0 Aug 80 1.50 
9    0 Sep 90 1.00 
10    1 Jan 10 1.50 
11    1 Feb 20 1.00 
12    1 Mar 30 1.00 
13    1 Apr 40 1.00 
14    1 May 50 1.50 
15    1 Jun 60 1.75 
16    1 Jul 70 1.75 
17    1 Aug 80 1.00 
18    1 Sep 90 1.00 

我想創建另一個數據幀，其中 - 每個代碼，我添加了前3個月 - 一月二月三月，加另一個3個月四月五月六月， 保持七月八月九月單獨

我的預期數據幀

  code   Month   Var1 var2 
1    0 <Jan+Feb+Mar> <10+20+30> <2.00 + 3.50 + 55.00> 
4    0 <Apr+May+Jun> <40+50+60> <23.50 + 1.00 + 1.00> 
7    0 Jul 70 1.00 
8    0 Aug 80 1.50 
9    0 Sep 90 1.00 
... similarly for code 2 

是否有實現這一目標的任何有效的方法？特別是使用使用data.tabledata.table

Answer 1

解決方案：

library(data.table) 
setDT(d) 
d[, group := Month] 
d[Month %in% c("Jan", "Feb", "Mar"), group := "Jan+Feb+Mar"] 
d[Month %in% c("Apr", "May", "Jun"), group := "Apr+May+Jun"] 
d[, .(Var1 = sum(Var1), var2 = sum(var2)), .(code, Month = group)] 

結果：

code  Month Var1 var2 
1: 0 Jan+Feb+Mar 60 60.50 
2: 0 Apr+May+Jun 150 25.50 
3: 0   Jul 70 1.00 
4: 0   Aug 80 1.50 
5: 0   Sep 90 1.00 
6: 1 Jan+Feb+Mar 60 3.50 
7: 1 Apr+May+Jun 150 4.25 
8: 1   Jul 70 1.75 
9: 1   Aug 80 1.00 
10: 1   Sep 90 1.00 

Answer 2

下面是創建使用該組的選項%/%

setDT(df1)[, grp := c('Jan_Feb_Mar', 'Apr_May_Jun')[(match(Month, month.abb)-1) %/% 3 + 1] 
      ][is.na(grp), grp := Month][, .(sum(var2)), .(code, grp)]