Release 2023-July

2023-July release notes for the Engine. This includes an integrated ML flow via our new interface, and improvements to the customer churn prediction template.

In this major release of July 2023, PI.EXCHANGE introduces users to an integrated ML flow to solve supervised ML problems. It provides seamless integration from data import, and preparation to model training and prediction generation. This would allow users to save a significant amount of time and effort in building their end-to-end ML pipeline. Additionally, we extend the capability of our customer churn prediction template to cover subscription-based use cases. Last but not least, the Engine now comes with improvements in setup stage for the detection and prevention of issues that potentially cause processing to fail, ensuring higher success rate of app creation.

Build from scratch

Understanding that the ML pipeline is comprised of simultaneous processes, we introduced an integrated ML flow via our new and improved interface. Using this interface, users can now easily build and use their ML pipeline with the following benefits:

  1. Import and prepare data then use it to train models in a single flow with significantly reduced processing wait time

  2. Automatically apply the same data preparation steps in model training to prediction generation to save time and effort

  3. Set up periodic prediction generation to automatically have prediction outputs available in your own database



To enhance the capability of the customer churn prediction template, the PI.EXCHANGE team has been diligently implementing the following improvements:

  1. On top of the transaction-based use cases, the customer churn prediction template also now supports subscription-based use cases to support more industries such as telecom, utilities and streaming services

  2. For both the transaction-based and subscription-based use cases, we have added some improvements to the process across the board, to increase the success rate of these apps.