How to use logistic model in WEKA Node?

Hi all,
I´m a new but impressed starter of KNIME. First results looking very interesting.

Here is one point where I need your help:
I downloaded the functionality of WEKA to KNIME. Here I tried to run a logistic model on my data
(you can find it under: WEKA-> functions-> logistic).
It worked fine when I use as a Input for the node a Test and Training data set containing the Target Variable. BUT: How can I use the trained model on data without a target variable. OR: How can I score data where the target variable is not included or unknown?

I would be happy for any suggestions or solutions.

Thank you very much.


Hello Thomas,

the test data is classified in the weka node, even if the target column is missing. There is a column for each possible class value containing the class probability.
However, there is a bug in the node that the winner class value is not appended, if the target column is missing in the test data. Thank you for this this remark, I will fix it for the next version of Knime, which is due to be released in a month.
IIn the meantime, you can either take a look at the class probabilites - or you use the joiner node (wih option 'Filter Duplicates'), connect its inports with the training and test data and connect the outport of the joiner node with the inport of the test data inport of the weka node. This produces a 'fake' class column, which makes the weka node produce the winner class column.
I hope this helps, please let me know.

- Nicolas

Hello Nicolas,

thank you very much for your reply. Your answer solved my problem. Everything is working properly now.