Chapter 8: Use Case – Customer Churn Prediction
In the last chapter, we discussed the alternatives to the Feast feature store available on the market. We looked at a few feature store offerings from cloud providers that are part of Machine Learning (ML) platform offerings, namely, SageMaker, Vertex AI, and Databricks. We also looked at a couple of other vendors that offer managed feature stores that can be used with your cloud provider, namely, Tecton and Hopsworks, of which Hopsworks is also open source. To get a feel for a managed feature store, we tried out an exercise on the SageMaker Feature Store and also briefly discussed ML best practices.
In this chapter, we will discuss an end-to-end use case of customer churn using a telecom dataset. We will walk through data cleaning, feature engineering, feature ingestion, model training, deployment, and monitoring. For this exercise, we will use a managed feature store – Amazon SageMaker. The reason for choosing SageMaker...