The TreeRank software permits to build tree-based ranking rules from data with binary labels according to recently developed learning algorithms (see the list of references below). It is one of the sole software that performs ROC optimization. It also allows for reading ranking trees (in regards to prediction), plotting the corresponding tree graphics and visualizing their statistical performance on training/test data samples (in the ROC or PR spaces) and outputs a list of graphics which facilitate much interpretation of the prediction rules produced. TreeRank also performs model selection, automatically or at hand, through an easy-to-use graphic interface.
Marine Depecker (Renault SA-Telecom ParisTech)
More from the Same Authors
2009 Poster: AUC optimization and the two-sample problem »
Stéphan Clémençon · Nicolas Vayatis · Marine Depecker