Optimal scaling trees

The framework of this paper is supervised statistical learning in data mining. A typical data-mining problem is to deal with large sets of within-groups correlated inputs compared to the number of observed objects. In case of complex relationships standard tree-based procedures offer unstable and not ever interpretable solutions. For that multiple splits defined upon a suitable combination of inputs are required. This paper provides a solution to build up a tree-based method whose nodes splitting is due to factorial multiple splitting categorical variables. A recursive partitioning algorithm is introduced considering a two-stage splitting criterion based on object scores from Nonlinear canonical correlation analysis.