In addition to this, a “black-box” method like random forests can be used to identify a small number of potentially relevant predictors from the full feature list, that can then be processed, e.g., by means of a familiar parametric method. This two-stage approach has been successfully applied in a variety of applications (see, e.g., Ward et al. 2006). Note, however, that variable selection should not be conducted before applying another statistical method on the same learning data (Ambroise and McLachlan 2002; Leeb and Pötscher 2006; Boulesteix et al. 2008).
