我有一个带有时间顺序索引的四列矩阵和三列名称(字符串)。这是一些玩具数据:
x = rbind(c(1,"sam","harry","joe"), c(2,"joe","sam","jack"),c(3,"jack","joe","jill"),c(4,"harry","jill","joe"))
我想创建三个额外的向量,以计数(每行)任何以前(但不是后续)名称的出现。这是玩具数据的所需结果:
y = rbind(c(0,0,0),c(1,1,0),c(1,2,0),c(1,1,3))
我将失去如何解决问题,并搜索了堆栈溢出以获取相关示例。Dplyr为查找总数提供了答案,但(据我所知)并非一排。
我试图编写一个函数以在单栏空间中处理此问题,但没有运气,即。
thing = sapply(x,function(i)length(grep(i,x[x[1:i]])))
任何提示都将不胜感激。
这是典型的ave
seq_along
问题类型,但是我们需要首先将数据转换为向量:
t(`dim<-`(ave(rep(1, prod(dim(x[, -1]))),
c(t(x[, -1])), FUN = seq_along) - 1,
rev(dim(x[, -1]))))
# [,1] [,2] [,3]
# [1,] 0 0 0
# [2,] 1 1 0
# [3,] 1 2 0
# [4,] 1 1 3
也许更可读性:
## x without the first column as a vector
x_vec <- c(t(x[, -1]))
## The values that you are looking to obtain...
y_vals <- ave(rep(1, length(x_vec)), x_vec, FUN = seq_along) - 1
## ... in the format you want to obtain them
matrix(y_vals, ncol = ncol(x) - 1, byrow = TRUE)
# [,1] [,2] [,3]
# [1,] 0 0 0
# [2,] 1 1 0
# [3,] 1 2 0
# [4,] 1 1 3
您可以做:
el = unique(c(x[,-1]))
val = Reduce(`+`, lapply(el, function(u) {b=c(t(x[,-1]))==u; b[b==T]=(cumsum(b[b==1])-1); b}))
matrix(val, ncol=ncol(x[,-1]), byrow=T)
# [,1] [,2] [,3]
#[1,] 0 0 0
#[2,] 1 1 0
#[3,] 1 2 0
#[4,] 1 1 3