Chapter 5: Model Training and Inference
In the last chapter, we discussed Feast deployment in the AWS cloud and set up S3 as an offline store and DynamoDB as an online store for the model. We also revisited the few stages of the ML life cycle using the Customer Lifetime Value (LTV/CLTV) model built in Chapter 1, An Overview of the Machine Learning Life Cycle. During the processing of model development, we performed data cleaning and feature engineering and produced the feature set for which the feature definitions were created and applied to Feast. In the end, we ingested the features into Feast successfully and we were also able to query the ingested data.
In this chapter, we will continue with the rest of the ML life cycle, which will involve model training, packaging, batch, and online model inference using the feature store. The goal of this chapter is to continue using the feature store infrastructure that was created in the previous chapter and go through the rest of the ML...