Regression Learner - Optimization the model

Hi All,

I have created a logistic regression workflow for classification problem. I want to improve the model accuracy which decided to use parameter optimization node to find the optimize paramters for the model.

In the regression learner, I decided to use SAG solver. Please advise which parameters (max. no of epoches, epilon or step size and variance) should be added in the optimization node. Thanks

hi @SIngpaore_knime ,

I found a reference on the knime hub that you can use to learn how to create a workflow for logistic regression parameter optimization. You can find the link here:

i’ve run this workflow with a small sample of data to give you an idea of the output.


I hope other data scientists and knime experts can also assist you.




Thank you for your advice.

I have added the Eplsilson m Var and stepsize in the optimization node but the logistics Regression learner node (Flow variables) did not allow me to link to the setting . The pull down manual pf epdilon, stepsize and priorVaraince are empty. I also attached my flow for all expect to advice . Many thanks

hi @SIngpaore_knime
i am not sure why the flow variables pull-down menu might not be enabled in your mentioned node. Could it be that this is happening only on this node? perhaps our knime expert may be able to troubleshoot and elaborate more on this issue. apologize for not being able to answer your question fully. i’m still learning, and not an expert in this area.


Thank you @marzukim. It is great you give me some direction.

Hi all expert,

Below is the warning message

WARN Parameter Optimization Loop Start 3:44:648 Unable to merge flow object stacks: Conflicting FlowObjects: <Loop Context (Head 3:44:648, Tail unassigned)> - iteration 0 vs. <Loop Context (Head 3:44:642, Tail unassigned)> - iteration 0 (loops/scopes not properly nested?)

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