Hello, I am trying to perform a Parameter Optimization over a Logistic Regression. I currently able to optimize over the numerical parameters eg. Max Epoches, Learning Rate or Prior Variance for Regularization. However I cannot figure out how to deploy an Optimization Parameter on variables such as Solver, Learning Rate Strategy and Regularization techniques that are “string” variables rather than numeric ones.
What do you suggest me?
Thank you in advance!
You could maybe use a numeric value and then a case switch / rule egine to choose this setting. My other option would be python
Could you please be a little more specific? If you have time can you please upload a possible python script as you suggested?
Sorry I don’t have the time right now. Maybe someone else can help. Have you tried an approach in knime with the case switch node. With python I would think of using a grid.
and welcome to KNIME Community!
Currently, the Parameter Optimization Loop Start node only supports numerical parameters. Check here for more:
How would I go about creating an optimization loop that chooses different categorical parameters? e.g. finding optimal kernel type in an SVM. It seems like you can only create parameters to search that are numerical.
I will add +1 on related ticket and someone will update ticket once there is news.
I have managed to solve my issue using Rule Engine Variable that allowed me to transform an integer into a string and then used in my algorithm.
Thank you anyway for the support!
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.