Output Decision Tree Predictor

Hey everyone,

I have a decision tree predictor knode that gives me the predicted values for my target variable. I also need a column that gives me the probability for these parametervalues to be true. Is there any chance I can get these probabilities in my ouput table?

Thanx, Martina

Currently the decision tree implementation only returns labels and no scores/confidences.


There is currently no way convincing the decision tree learner to produce confidence values.

The plans for 2.0 are that we will modify all of our learner (and predictor) nodes to use dedicated port types (that has different reasons: type checking at "connection time", configuring predictors without having to execute the learner node first, memory efficiency, ...).

As part of that process we also plan to provide those nodes (where applicable) with additional confidence columns. And yes, the decision tree learner is a good candidate for that. Same is true for all weka classifiers (that has already been done, they are just not yet released.)