Classifying pairs with trees for supervised biological network inference† †Electronic supplementary information (ESI) available: Implementation and computational issues, supplementary performance curves, and illustration of interpretability of trees. See DOI: 10.1039/c5mb00174a Click here for additional data file.

3 Illustration of interpretability of trees 7 3.1 Interpretability of single decision trees . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.1.1 Local approach with two multiple-output trees . . . . . . . . . . . . . . . . . . . 9 3.1.2 Global approach with one single-output tree . . . . . . . . . . . . . . . . . . . . 12 3.2 Clustering with ensembles of trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.2.1 Partitioning the matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.2.2 Characterizing the regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.2.3 Network of features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.3 Pair-based feature ranking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20