Sebastian Fischer

Stuff that is often wrong / suboptimal: * crating the custom_check function * default vs initial parameters * ordered features (?) * ...

There seems to be some progress to make catboost available as an R pacakge: https://github.com/catboost/catboost/issues/439 This issue is merely a reminder to use the CRAN package once this happens

```r des = data.frame(min.bucket = grid, min.node.size = grid) tuner = tnr("design_points", design = des) ``` cc @berndbischl