We would like to know if KNIME provides flexibility to modify the parameter estimate of the variable, once the model has been developed by KNIME engine. Can we do parameter tuning / tweak parameter estimate after model development stage?
This is specific to modifying parameter estimates (model development stage) and not hyper parameter tuning…just a fyi…
At first glance, my initial answer to your question would be no, apart from manually editing XML rules generated in a PMML decision tree, for example. But I am probably missing some context or nuance in what you’re trying to do.
Is there a specific use case you’re working on here that you could provide some more context about?
Thanks for your reply, we are trying to evaluate that if KNIME engine gives us option to modify the parameters, once the model has been built and re-run the model again on training data once the parameters has been modified.
Lets say linear model developed by KNIME is y~ax+bz+c. My question is, can we modify parameters a & b on the fly and rerun the model on training dataset to see how that impacts my model accuracy or other factors.
I guess it is possible to build such a workflow (depending on the model) but to my knowledge KNIME doesn’t offer this out of box. But way such approach? Usually configuration parameters are modified and optimized and not result parameters…