Automated Self Learning workflow

Hello,

I am looking to accomplish a workflow that looks to compare existing model with a new model. The completed workflow will be able to ingest data and run it on different models and the winning model’s PMML is outputed. Then I upload another dataset which has some numeric changes to the workflow and its able to run the new data on previously successful model and also compare it with the other models and outputs the result. What I am trying to accomplish here is to be able to build a workflow that will be able to take new data and compare it with the champion model and tell me if the champion is good or if a different model performs better. Any help appreciated.

Sid

You could try and explore the KNIME model factory concept and see if you could adapt it.

https://hub.knime.com/knime/spaces/Examples/latest/50_Applications/26_Model_Process_Management/01_Model_Factory

Also note the links below the workflow picture.

2 Likes

Thanks for pointing out the Model Factory workflow, I tried to deploy the whole workflow to the server and run it but am not able to. I took care to upload the supporting folders as well with the workflow. Do I need to do anything here before deploying ?

I think you might want to read the accompanying articles by @Iris and others. And you might have to adapt the concept to your needs.

I would have to take a closer look myself.

Hi @sramesh -

After looking at this over the weekend, I noticed there is a problem with running the model factory as-is. We’re looking at it, and I’ll be sure to let you know when it’s been fixed. Sorry for the trouble.

1 Like

Thanks for letting me know. Is there any config changes that I could make changes in the config of the workflow and get it to work. I am working on a POC and the model factory is an imp piece for that and it would be great to get to run it on the server.

This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.