Reputation: 67
If these are the first 6 rows of my nx6 matrix
x1 x2 x3 x4 x5 a1
1 30 3 43 0 0.20
3 40 2 35 0 0.21
3 50 7 72 0 0.75
4 40 58 63 10 0.55
1 20 19 61 10 0.43
4 20 5 49 0 0.62
The goal is to create a new nx5
matrix based on this logic. Each row from X1-X5 should be multiplied by the scalar values in a1 like this
a2
[1,30,3,43,0]*0.20 + [1,30,3,43,0]*0.21 + [1,30,3,43,0]*0.75 + [1,30,3,43,0]*0.55 + [1,30,3,43,0]*0.43 + [1,30,3,43,0]*0.62
[3,40,2,35,0]*0.21 + [3,40,2,35,0]*0.75 + [3,40,2,35,0]*0.55 + [3,40,2,35,0]*0.43 + [3,40,2,35,0]*0.62
[3,50,7,72,0]*0.75 + [3,50,7,72,0]*0.55 + [3,50,7,72,0]*0.43 + [3,50,7,72,0]*0.62
[4,40,58,63,10]*0.55 + [4,40,58,63,10]*0.43 + [4,40,58,63,10]*0.62
[1,20,19,61,10]*0.43 + [1,20,19,61,10]*0.62
[4,20,5,49,0]*0.62
Finally resulting is a nx5 matrix below
2.76 82.8 8.28 118.68 0
7.68 102.4 5.12 89.6 0
7.05 117.5 16.45 169.2 0
6.4 64 92.8 100.8 16
1.05 21 19.95 64.05 10.5
2.48 12.4 3.1 30.38 0
Any pointers for accomplishing this appreciated. Thanks.
Upvotes: 1
Views: 75
Reputation: 886938
We could loop over the sequence of rows with sapply
, then do a the crossprod between the last column and the rest while we use the elements from that row to the last for the last column
t( sapply(seq_len(nrow(m1)), function(i) {
x1 <- m1[i:nrow(m1),6]
x2 <- m1[i, -6]
matrix(x2, length(x2), length(x1)) %*% x1 }))
m1 <- structure(c(1, 3, 3, 4, 1, 4, 30, 40, 50, 40, 20, 20, 3, 2, 7,
58, 19, 5, 43, 35, 72, 63, 61, 49, 0, 0, 0, 10, 10, 0, 0.2, 0.21,
0.75, 0.55, 0.43, 0.62), .Dim = c(6L, 6L), .Dimnames = list(NULL,
c("x1", "x2", "x3", "x4", "x5", "a1")))
Upvotes: 1