Regression model which takes Fingerprints


Are there any plans to introduce a new node in the Statistics/Regression section which can take chemical fingerprints as input to generate a regression model predictor.

This seems somewhat of a missing feature in KNIME, as currently there is only the Fingerprint Bayesian Classification Model.


Hi Simon, 

I've logged this request in our issue tracking system.  FYI, the naive bayes fingerprint nodes were sponsored development project.  I'll look into the feasiblity of extending this feature set to include regression nodes as well. 

Regards, Aaron

Many thanks Aaron. Incorporating a regression node would be extremely beneficial to all pharma colleagues in generating various predictive models.




Hi Simon,

The Tree Ensemble Learner (Regression)/Predictor also has regression variant and accepts molecular fingerprints. The description also refers to the classification learner, which is described here (and has other pointers to wikipedia).


Hey bernd,
Thanks for pointing me to this, this is perfect, just what I need, and it seems to be working pretty well.