Analyse Node for Decission Tree

KNIME Analytics Platform
1

Hello,

i need an analysing node for my decission tree.

An example: I have got a data set about 200 patients that gets different drugs depend on their health status. I've got a "Scorer" Node with the confusion Matrix...its behind the "Decission Tree Predictor" and "Decission Tree Learner" node.

The results from the confusion Matrix shows defference between KNIME and Clementine:
KNIME show that all predicted values are correct, and Clementine that one data is wrong.

What i want is an analysis node for the validiation of my decission tree according to see false predicted values.

Are there possibilities to do that?

greetz

2

Nice to see that the KNIME decision tree outperforms others :-)

More seriously - I guess this has to do with different settings in the learning
algorithms. You can put the training data into the decision tree predictor node
(and connect this also to the trained decision tree) and then look at the rows in
your output table which have different predicted classes.

But in your particular example, this will not be the case for the any of the rows,
of course.

Or am I misreading what you are trying to do?

- Michael