Hi,
Is it possible to convert a Tree Ensemble model to a PMML model? I got some hints from a blog (http://www.dataminingreporting.com/blog/-export-random-forest-as-pmml), to first extract the model with the Tree Ensemble Model Extract node, followed by the Cell to PMML node (see image). However, while the Tree Ensemble model consists of 100 trees, the Cell to PMML node only converts the first row, so I would think that the PMML model cannot be complete. This is supported by the fact that the predictive power of the PMML model is lower than the original model, when predicting the training set.
Also, can PMML models handle RDKit fingerprints? I get an error message from the PMML Predictor node if I don't expand the fingerprints first.
Thanks,
Evert