[R] caret for non-numeric data

[Yang Zhang]

caret has many useful preprocessing facilities for numeric predictors,
such as removing near-zero-variance predictors, removing correlated
predictors, removing linear combinations, centering/scaling, and
Box-Cox transformations.  However, these don't work for factor data,
which is common in many learning tasks.  Is there anything similar for
factor data?  Thanks in advance.