R integration


Awhile back it was mentioned that a higher degree of integration with R was being looked into? How is that progressing?

This is a very interesting feature as it makes Knime/R kind of like Insightful Miner/S-Plus. I have been experimenting with the existing capabilities of Knime more over the last month and have grown to like the product even more!

Best Regards,


we are working on this - v1.3 will have better R-integration so that one does not need to rely on the flaky Rserve setup anymore.

Stay tuned.
- Michael

Excellent thanks Michael!


I would be happy to help with testing if you think another set of eyes would be helpful here. Let me know.

Best Regards,


As long as you are improving the R integration, I would like to suggest a feature. I would like to be able to implement a learner and predictor in R. This would require being able to output a model from a node. It would also require a two input node, one input of which can be the model created from a previous R node. This would really be a cool feature.

1.3 didn't really bring the desired change, did it?

EDIT: Never mind, I just checked the release notes for 1.3.1, stating that the local execution is in fact "in" since 1.3, regardless of the R extension download page still pretending to require RServe. Could you correct that, please?

Can we expect a simple "unzip and forget" R node anytime soon, maybe in 2.0? I'm not talking of getting rid of the R server alone, but also of having to install R as such... :-)

EDIT: According to this post, R as such is perfectly portable... Maybe I'll package this node myself, even though it obviously won't serve as an excuse towards my rather restrictive IT department.


Hi Ergonomist,

thank you very much for your comments! We've updated the R extension download page.
As you are completely right - since 1.3 R nodes within KNIME also run with a local instance of R and therefore without the R Server. Again, thank you for the hint!


Thanks (as always) for the quick reaction! :-)