hyperparameter tuning for string values

KNIME Analytics Platform
1

Hi there,
I read the forum post about ‘hyperparameter tuning for string values’ (Hyperparameter tuning - #2 by julian.bunzel). I couldn’t find any other post on this. Is an extension (for strings) of the “parameter optimization loop” node a topic for further development?

thx
Ralf

2

Hello @ralf42 -

We do still have a ticket open for this (AP-11344) but I don’t have a timeline for you on when that might be implemented. I added you as a +1 to the ticket. Thanks for asking!

1 Like
3

@ralf42 you could revert to using (eg.) H2O.ai with the KNIME Python nodes which does support string Hyper Parameters like “categorical_encoding”. I have a Jupyter notebook somewhere with that and could try to pack that into a KNIME Python node.

https://docs.h2o.ai/h2o/latest-stable/h2o-docs/grid-search.html

4

@mlauber71 thx for your offer. I already have a solution based on python/sklearn/RandomizedSearchCV or GridSearchCV. I was looking for a solution in the knime-standard without an additional python-node.

@ScottF: thx for your reply - so i will wait.

Best
Ralf

1 Like
5

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.