Hamilton Open Source

As I understand both `age_mean` i `age_std_dev` are part of model training - you calculate those values from the data itself.  I would keep them as part of the model.

The tricky part is when you calculate features in realtime from realtime events.

Scenario I have in mind is the following (this is a real life scenario):
1. User selects objects into a set.
2. Set is put into a transformation function to calculate features.
3. Features from 2 are concatenated from batch features
4. Features from 2 and 3 are fed into an ML model and the score is calculated
5. User checks score and depending on the value jump to 1 or 6
6. end


In this scenario you calculate features for dynamically created sets of objects. No, known to me, feature store can handle this easily.

But good news is that Hamilton can handle this easily :slightly_smiling_face:

Yeah great. Thanks for the thoughts <@U048HD29YP5>.
&gt;  I would keep them as part of the model.
yeah I’m trying to keep the software engineering to a minimum with the example, but I want people to know that’s a reasonable option. :thinking_face: