Martin Martin - 13 days ago 7
R Question

R: Renaming when using model.matrix

When running the following R code:

y <- rnorm(100)
B <- matrix(rnorm(200), ncol=2)
colnames(B) <- c("x1", "x2")
A <- model.matrix(y~B)
colnames(A)


model.matrix adds to the variable names the prefix "B" in the matrix A. How can I avoid this behavior (or rename it but in a very general setting, e.g. within a function with user supplied formula)?

Answer

We can use sub to remove the first character in the column names.

colnames(A)[-1] <- sub("^.", "", colnames(A)[-1])

Or set the column names by concatenating with the ones from 'B'

colnames(A) <- c(colnames(A)[1], colnames(B))

Another option without using the sub would be to create a data.frame with 'y' and 'B' matrix and then use y ~.

A <- model.matrix(y~., data=data.frame(y, B))
colnames(A)
#[1] "(Intercept)" "x1"          "x2"