Tree Ensamble Learner (Regression) node odd behavior (bug?)

Dear KNIMErs,

I'm using the Tree Ensamble Learner (Regression) node in the KNIME Labs repo and, in contrast with the similar classification node, I'm experiencing an odd behavior. I'm trying to set it to predict a response using a fingerprint bit vector column.

If in the input table are present only the response and the fingerprint columns the node configure dialog cannot be opened and the following message is obtained:

The dialog cannot be opened for the following reason: 

No column in spec compatible to "NominalValue" or "DoubleValue".

Anyway (oddly) the node can be executed (supposedly using the default configuration).

However if there is even a single one more column in the input table, than the node configure dialog can be opened and properly configured.

Please can anybody reproduce this behavior and/or confirm it is a bug?

Thanks in advance,


Hi Gio,

Yes, confirmed. It's a bug in the configuration dialog. Workaround is obviously to add a fake column and then ignore it as part of the configuration.

The problem will be fixed in 2.11.1.


OK Bernd, very good.

I'm already using it with the workaround but I wanted to post this for you (KNIME people) to be aware of it and fix it in the next releases. Thanks!