Fig. 5
From: PROMO: an interactive tool for analyzing clinically-labeled multi-omic cancer datasets
Automatically generated decision tree for classifying breast tumors into the four PAM50 classes. PROMO can generate a cross-validated decision tree for any selected sample label using the currently loaded matrix as training data. In this figure, a four-gene molecular classifier for breast cancer subtypes is presented, showing a 7.77% loss on the training data, and a 15% averaged loss on 10-fold cross-validation