Hello everyone,
We are experimenting with several approaches in a Machine Learning project ( binary classification), and we would like to keep track of those using MLFlow. We are using the feature store to build, store, and retrieve the features, and h2o to do the modeling. The approaches we are trying involve combinations of the following:
- Changing the features used
- Filtering the dataset (keeping or discarding certain records)
I have yet to find a way to keep track of those things in an organized way in MLFlow, except for writing the information somehow in the run description, but that does not seem right. I have also tried to write it as a parameter, but for instance the feature list exceeds the size limit. Is there a way to do this "correctly"?
Thank you
