如何创建标准尺寸的向量，以便可以将它们添加到数据框中？

Question

I have a single vector (call it t1) with a series of observations. I want to create a set of new vectors by popping the first observation from t1 (and so on for subsequent near-copies). But I want to keep the vectors the same length so I can add them to a data frame later.

I was able to make it work as follows:

t1 <- c(1, 2, 3)
t2 <- t1[-1]
t3 <- t2[-1]

t2[length(t2)+1] <- 0

t3[length(t3)+1] <- 0
t3[length(t3)+1] <- 0

t.all <- cbind(as.data.frame(t1), as.data.frame(t2), as.data.frame(t3))

t.all

  t1 t2 t3
1  1  2  3
2  2  3  0
3  3  0  0

But this is clumsy and it's going to be tedious if I want to create a large number of columns. How can I keep the vectors the same length (or solve this problem another way)?

Answer 1

Here a loop version of what you try to do , uding do.call and lapply :

cbind(t1,do.call(cbind,lapply(seq_along(t1)-1,
                     function(x)c(tail(t1,-x),rep(0,x)))))


    t1    
[1,]  1 2 3
[2,]  2 3 0
[3,]  3 0 0

Answer 2

Here's another way using vector indexing:

t1 <- (2,5,3)
mm <- do.call(rbind, lapply(seq_along(t1), function(x) t1[x:length(t1)][1:length(t1)]))
#      [,1] [,2] [,3]
# [1,]    2    5    3
# [2,]    5    3   NA
# [3,]    3   NA   NA

mm[is.na(mm)] <- 0
#      [,1] [,2] [,3]
# [1,]    2    5    3
# [2,]    5    3    0
# [3,]    3    0    0

Another way without using apply family:

t1 <- c(2,5,4,6)
len <- length(t1)
matrix(t1[outer(1:len, 0:(len-1), '+')], ncol=len)

#      [,1] [,2] [,3] [,4]
# [1,]    2    5    4    6
# [2,]    5    4    6   NA
# [3,]    4    6   NA   NA
# [4,]    6   NA   NA   NA

Answer 3

> t.all <- sapply(0:2, function(x) c( t1[(x+1):3], rep(0,x) ) )
> t.all
     [,1] [,2] [,3]
[1,]    1    2    3
[2,]    2    3    0
[3,]    3    0    0

If you need it to be a data.frame it would be a lot more efficient to build as a matrix first and then wrap as.data.frame around the final result.

Answer 4

How about creating a matrix row-by-row, by recycling t1 as desired:

tmat <-cbind(t1,t1,t1,t1,....) # as many as needed

Then just use a matrix triangle function

newmat<- tmat * upper.tri(tmat,diag=TRUE) 

That's offset from your sample, but contains the same info per row.

Answer 5

Most of the other answers focus on creating the final data.frame. If that is your ultimate goal, then they provide good approaches. This answer instead focuses narrowly on your question of how to take the first element off and preserve the length. In order to keep things tidy, it is best to do the whole thing in one function.

shift <- function(tx) {append(tx[-1],0)}

Then you can have

t1 <- c(1, 2, 3)
t2 <- shift(t1)
t3 <- shift(t2)

t.all <- data.frame(t1, t2, t3)

which gives you the same result you had.

> t.all
  t1 t2 t3
1  1  2  3
2  2  3  0
3  3  0  0

If you want to combine this function with a looping construct to create the data.frame, it is easiest to go through a matrix first.

t.all <- matrix(t1, nrow=length(t1), ncol=length(t1))
lapply(seq(length=length(t1))[-1], function(i) {
  t.all[,i] <<- shift(t.all[,(i-1)])
})
t.all <- as.data.frame(t.all)

which gives the same data.frame, but with slightly different column names

> t.all
  V1 V2 V3
1  1  2  3
2  2  3  0
3  3  0  0