What's next

Part 1 complete

You now have a working feature-store backbone in Aerospike: feature-group and feature metadata define what each signal means, entity records store model-ready values by key, and dataset definitions capture reproducible training slices. That gives training and serving a shared foundation so you can move into model development without rebuilding your data model.

In Part 2, you will create a new synthetic dataset for model development and store its feature definitions and entity records in Aerospike. You’ll then use the Dataset class to define what data to include in the training slice, then actually train and test the model using Spark MLlib.

Finally, in Part 3, you will finish the ML serving path by using Aerospike’s Python client to retrieve features and run inference.

Continue to Part 2: Model Training to train your first decline-risk model with this same feature store.

Checklist

• I have a working feature store ready for Part 2.

Share

Congratulations on completing this tutorial! Share your achievement with the world and let everyone know about your newly acquired skills.

Next Steps

• Continue to Part 2: Model Training to train ML models using feature store data.

• Explore the Aerospike Spark connector documentation for advanced features.