The statement about coercing to factor b/c of matrix requirements is simply wrong. (R matrices are incapable of holding factor variables.) Perhaps you are thinking of a data.frame. As the R FAQ says you need to use:
go$V4 <- as.numeric(as. character(go$V4))
If a numeric vector is concatenated (with c()
) to any character vector, it is immediately coerced to "character" mode. If a column in a text field has non-numeric characters in it, the same thing happens on input.