Is there a way to ensure that model matrix has the same dimensions for train and test set?
I tried the following:
library(tidyverse) create_encoder &l