More Math #41
Description
Problem
We are currently calculating a number of metrics from both individual vectors as well as populations across the training and rolling data sets. There is room for improvement, however, in that there are additional drift measures such as KL divergence and Earth Movers distance that would be useful to calculate in this workflow.
Solution
It is our opinion that either of these could be added to the tkyoDrift analysis with minimal effort, as the tkyoDrift main logic tkyoDriftSetTraining.py batch embedding logic both capture scalar metrics for population comparisons already.
Additional information
No response
👨👧👦 Contributing
- 🙋♂️ Yes, I'd love to make a PR to implement this feature!