Updating active models in predictions through API with MLOps
You can create and deploy models directly from your modeling tool to Pega Platform, by using a scripting language of your choice and the V2 Prediction API endpoints. The API provides you with options to remotely perform Machine Learning Operations (MLOps): add a model to your application, review (approve or reject) a model update, and retrieve the status of a model update.
Use this API to integrate the model approval process in Pega Platform with any external model deployment process that your organization uses.
To learn about the work flows, components, and architecture behind model updates, see Understanding MLOps.
V2 Prediction API service package
The V2 Prediction API service package, which you can access from Dev Studio, describes the model update endpoints, requests, and responses. Use the service package to test the endpoints by generating a sample response to a simulated endpoint. For more information, see Accessing and testing V2 Prediction API.
MLOps prerequisites
To replace models in predictions through the V2 Prediction API, you need to configure your Business Change pipeline, Prediction Studio settings, and data scientist operators to meet the requirements for the model update feature. For more information, see MLOps prerequisites.
Model update endpoints
Learn more about the model update endpoints by reviewing sample responses and parameter descriptions:
- Starting a model update
Use this endpoint to start an update process to replace a model in a prediction by using Machine Learning Operations (MLOps).
- Reviewing a model update
Use this endpoint to approve or reject a candidate model in a prediction by using Machine Learning Operations (MLOps).
- Retrieving the model update status
Use this endpoint to load the status of a model update that was started by using Machine Learning Operations (MLOps).
Previous topic Rejecting shadow models with MLOps Next topic Starting a model update