Add optional Feast feature store notebook and integration cells#135
Add optional Feast feature store notebook and integration cells#135aniketpalu wants to merge 1 commit intorh-aiservices-bu:mainfrom
Conversation
- New notebook 8_feature_store_feast.ipynb: self-contained Feast tutorial covering feature definitions, on-demand features, historical/online retrieval, data integrity verification, and production deployment guide - Append optional Feast cells to 1_experiment_train.ipynb (get_historical_features) - Append optional Feast cells to 3_rest_requests.ipynb (get_online_features) - All Feast cells are try/except guarded — zero impact on existing workshop flow - Uses local Feast mode (file registry, SQLite) — no operator/pods required
|
@RHRolun @MelissaFlinn 👋 Feature Store engineering team worked on adding the capability to the Fraud Detection tutorial. The idea is to show the value of a feature store. There are still a couple of feedback that the team needs to address, but would you mind looking at it and giving your initial feedback/concerns/change requests, if any 💟 Thank you so much!! |
|
a feedback from me: would it be better to keep the first notebooks as it, and have additional notebooks like "8.1-experiment_train_with_feast.ipynb", " 8.2-rest_request_with_feast.ipynb" Because I think it would be better letting people to go through the process, the basic/classic data science process. And then show introduce Feast, and show them doing them with Feast while iterating the value of Feast can be better. It would allow people to compare the two approaches (with Feast vs without Feast) what do you think @RHRolun? 🙏 |
2 participants
Summary
What the notebook covers