R: How to rescale my matrix by column

Question

I have a matrix of disease states states0CommercialA, where the columns are states (i.e., "no disease", "disease", "dead") and the rows are model cycles (i.e., 1, 2, 3, 4, etc.).

I'm trying to multiply this by a vector of costs commercialMedExp, wherein each cost corresponds to a disease state. I've attempted the following:

commercialMedExp * states0CommercialA

but it appears as though the multiplication is occurring across columns instead of across rows. Could someone help me with the correct syntax?

Answer 1

Something like

commercialMedExp0A <- t(apply(states0CommercialA, 1, function(x){ x * commercialMedExp}))

should work so long as the number of columns in states0CommericialA is the same length as commercialMedExp. If it is not you would have to subset the data. For example, if the disease states are in columns 13 through 18

    commercialMedExp0A <- t(apply(states0CommercialA[,c(13:18)], 1, function(x){ x * commercialMedExp}))

Answer 2

Column / Row rescaling is a common operation in matrix computation. You are looking for column rescaling, but I will offer solutions to both.

---

Row rescaling

A <- matrix(1:20, nrow = 5); x <- 1:5
## Any method below is much more efficient than `diag(x) %*% A`

## method 1: recycling
A * x

## method 2: `sweep()`
sweep(A, 1L, x, "*")

Column rescaling

A <- matrix(1:20, nrow = 5); y <- 1:4
## Any below method is much more efficient than `A %*% diag(y)`

## method 1: transpose + row rescaling
t(y * t(A))

## method 2: `sweep()`
sweep(A, 2L, y, "*")

## method 3: pairwise multiplication
A * rep(y, each = nrow(A))

---

What you can do

states0CommercialA * rep(commercialMedExp, each = nrow(states0CommercialA))