I work on decision-tree methods a lot, because single decision trees are easy to understand by non-statisticians. With ensembling techniques like gradient boosting, trees can provide state-of-the-art predictive accuracy. Here is a short interview about my work on trees (also in Dutch). My current focus is in combining trees with mixed-effects models, smoothing splines and generalized additive models.
I develop open-source software packages for the programming environment R (see left or GitHub). My work has been published in journals like Psychological Methods, the Journal of Statistical Software, Information Fusion and World Psychiatry (see menu on the left or Google Scholar).
As Mary-Jo and the Support Vector Machines, I make music videos about statistical methods (see menu on the left or YouTube).