R數據幀度量到數組維度

Question

我有這個data.frame。我想將類（c1到c3）轉換爲維度和度量（m1,m2）到另一個維度。

df 
    c1 c2 c3  m1  m2 
1 0 FALSE C 0.4572084 26.453 
2 0 FALSE V 0.4033657 11.505 
3 0 TRUE C  NA  NA 
4 0 TRUE V 0.4665911 23.062 
5 8 FALSE C 0.4566605 27.685 
6 8 FALSE V 0.3920727 3.505 
7 8 TRUE C  NA  NA 
8 8 TRUE V 0.4528438 16.155 
9 2 FALSE C 0.4401774 29.609 
10 2 FALSE V 0.4161140 10.410 
11 2 TRUE C 0.3979405 5.840 
12 2 TRUE V 0.4268235 10.168 

它可以這樣做：

library(reshape2) 
a=acast(df,c1~c2~c3,value.var = 'm1',fun.aggregate=mean) 
b=acast(df,c1~c2~c3,value.var = 'm2',fun.aggregate=mean) 
c=array(c(a,b),dim=c(3,2,2,2)) 
dimnames(c)=list(c('0','2','8'),c('F','T'),c('C','V'),c('m1','m2')) 
c 
, , C, m1 

      F   T 
0 0.4572084  NA 
2 0.4401774 0.3979405 
8 0.4566605  NA 

, , V, m1 

      F   T 
0 0.4033657 0.4665911 
2 0.4161140 0.4268235 
8 0.3920727 0.4528438 

, , C, m2 

     F T 
0 26.453 NA 
2 29.609 5.84 
8 27.685 NA 

, , V, m2 

     F  T 
0 11.505 23.062 
2 10.410 10.168 
8 3.505 16.155 

有沒有辦法一步到位做到這一點？
我的最終目標是要做到這一點：

c['8','F','V','m2'] 
[1] 3.505 

的另一種方式做到這一點，不使用數組，是根據級聯的ID來命名行。

rownames(df) = paste0(df$c1, substr(df$c2,1,1), df$c3) 
df['8FV','m2'] 
[1] 3.505 

---

df <- structure(list(c1 = c(0L, 0L, 0L, 0L, 8L, 8L, 8L, 8L, 2L, 2L, 
2L, 2L), c2 = c(FALSE, FALSE, TRUE, TRUE, FALSE, FALSE, TRUE, 
TRUE, FALSE, FALSE, TRUE, TRUE), c3 = c("C", "V", "C", "V", "C", 
"V", "C", "V", "C", "V", "C", "V"), m1 = c(0.4572084, 0.4033657, 
NA, 0.4665911, 0.4566605, 0.3920727, NA, 0.4528438, 0.4401774, 
0.416114, 0.3979405, 0.4268235), m2 = c(26.453, 11.505, NA, 23.062, 
27.685, 3.505, NA, 16.155, 29.609, 10.41, 5.84, 10.168)), .Names = c("c1", 
"c2", "c3", "m1", "m2"), class = "data.frame", row.names = c("1", 
"2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12")) 

Answer 1

我想我找到了你一個更簡單的選擇：

library(reshape2) 
m <- melt(df, id.vars = c("c1", "c2", "c3")) 
a <- acast(m,formula = c1 ~ c2 ~ c3 ~ variable) 
a['8','FALSE','V','m2'] 
[1] 3.505