CI/CD Pipeline

In this release:

• Integration with CodeDeploy and GitHub for a full CI/CD Pipeline
• As part of this preprocessor storing, training, and tuning in the build phase
• Initialisation now down in .env, not through key word arguments in bash
• One click deployment for inference after the build in CodeDeploy

All thanks to great work of @inchara1990 💯 🎉

*The previous version is stored in branch main-oldversion reference.

Customer Segementation

This release introduces the option to select customer segmentation by passing parameter inside the cloud formation template.
If cluster=True, the during preprocessing the data (regardless of data type) will be clustered using Amazon DenseClus Package.

This allows for clusters/segments to later be used for marketing and communication strategy.

Churn Time to Event

This release introduces using Time to Event modeling for Churn using Cox Proportional Hazards.

Includes separates scripts and CloudFormation template for running - along with demonstration notebook.

Documentation Release

Added documentation and docstrings.
Added parameters for inference preprocessing and evaluation.

v0.1

First release of the churn pipeline.

Includes:

data ingestion, training/retraining, and inference by using Amazon Lambda calls to trigger pipeline runs.

Allows for the ability to automate runs, experiment rapidly, deploy quickly and easily incorporate expanding data sizes at the Tera-byte level.