### conditional inference trees

o  ctree(y ~ x1 + x2 - x3) will not work

o  implement quantile regression (Meinshausen, JMLR)

o  implement cvrisk for ctree and cforest (now that predictions for obs with zero

   weight are computed)

o  we can re-compute prediction weights from the weights in the root

   node, the terminal node number and where.

o  more checks for MonteCarlo

o  check convenience functions (`treepredict' seems the

   wrong name)

o  predictions for multiple response including factors

o  check storage mode of ordinal variables and fix varimp

o  missing values and splitting: weights isn't adjusted for functions in

   Split.c

o  implement splits in binary factors without looking at statistics (check

   only minsplit)

o  vignette for cforest

o  varimp for censored observations

o  varimp for matrix interface

### model-based trees

o  better docu for mob (incl. journal + survreg)