Some questions about C4.5 tree


Is there a way to get or to generate probability values for C4.5 tree predictor in order to use ROC curve node? Also as far as I understood J48 is weka implementation of C4.5 but with default settings of the learner nodes predictions and statistic are different. Is this normal?


The Decision Tree Predictor has an option "Append column with normalized class distribution" - is this what you are looking for? The KNIME learner is based on the SPRINT implementation that may explain the differences.