Drop columns by name data frame vs matrix

Question

I want to drop columns by name in a matrix, I noticed that it does not work the same as for data frame:

df <- as.matrix(data.frame(x=1:5, y=2:6, z=3:7, u=4:8))

df[ , -which(names(df) %in% c("z","u"))]


df <- data.frame(x=1:5, y=2:6, z=3:7, u=4:8)

df[ , -which(names(df) %in% c("z","u"))]

Why and how can I fix this?

Because matrices and data frames are different. Columns aren't as special in matrices as they are in data frames. `colnames` will work for both. — Gregor Thomas, Oct 13 '17 at 15:46
`data.frame` is essentially `list` with properties of matrix in R. You can't expect matrix to be having every method same as `data.frame`. — platypus, Oct 13 '17 at 15:50

score 1 · Accepted Answer · answered Oct 13 '17 at 16:03

With a matrix you can use colnames or rownames (or, if you need to generalize up to an array with more dimensions, the dimnames list).

names isn't defined for a matrix. It is defined for a list, and thus for a data.frame, where columns are generally more important (in some sense, at least), so it is a safe convention that names refers to column names. But in an array there is no reason to prefer one dimension to another.

A similar question is Extract matrix column values by matrix column name.

Drop columns by name data frame vs matrix

1 Answers1