R如何合并不同的数据帧列

Question

我在R中有一个数据框,看起来像这样

data
x1  x2  x3a x3b x3c x3d x4
A   43  0   0   0   1   P
B   23  0   1   0   0   Q
C   11  0   0   0   0   R 
D   66  0   0   1   0   S

现在我想将列合并x3a, x3b, x3c, x3d到单列.预期的单列将包含值为1的列号x3a,x3b,x3c,x3d.该值应编号(x3a=1,x3b=2,x3c=3,x3d=4).期待结果如下

x3
[1] 4    2    0    3

我试过reshape()功能,但无法得到我真正想要的东西

q<-data[,3:6]
r<-reshape(q,varying=c("x3a","x3b","x3c","x3d"),v.names="x3",direction="long",times=c("x3a","x3b","x3c","x3d"))
final<-r[r$x3!=0,][,3]

但这没有给出预期的结果.它错过了价值0之间2和3:

final
[1]4    2    3 

Answer 1

这有效:

data <- data.frame(
  x1 = c('A','B','C','D'),
  x2 = c(43,23,11,66),
  x3a = c(0,0,0,0),
  x3b = c(0,1,0,0),
  x3c = c(0,0,0,1),
  x3d = c(1,0,0,0),
  x4 = c('P','Q','R','S')
)
data$x3 <- as.matrix(data[,c('x3a','x3b','x3c','x3d')]) %*% c(1,2,3,4)

结果:

  x1 x2 x3a x3b x3c x3d x4 x3
1  A 43   0   0   0   1  P  4
2  B 23   0   1   0   0  Q  2
3  C 11   0   0   0   0  R  0
4  D 66   0   0   1   0  S  3

大通发表了相关评论:如果x3a ... x3d不同于零或一个怎么办？您可以使用ifelse()以考虑该场景:

data$x3 <- as.matrix(ifelse(data[,c('x3a','x3b','x3c','x3d')] > 0, 1, 0)) %*% c(1,2,3,4)

Answer 2

@Barrankas的答案很聪明,也有矢量化,这里是一个不太聪明/矢量化的选项

as.numeric(apply(data[, 3:6], 1, function(x) which(x == 1)))
## [1]  4  2 NA  3

Answer 3

使用row和col索引.应该很快,因为你只分配一次.

data$new <- 0
tmp <- data[3:6]==1
data$new[ row(tmp)[tmp] ] <- col(tmp)[tmp]

data

#  x1 x2 x3a x3b x3c x3d x4 new
#1  A 43   0   0   0   1  P   4
#2  B 23   0   1   0   0  Q   2
#3  C 11   0   0   0   0  R   0
#4  D 66   0   0   1   0  S   3

tmp 可以更改以适应任何需要的逻辑比较.