At SAP TechEd 2017, I delivered a session on Machine Learning in the Cloud which comprised of details around SAP Predictive Services and Predictive Analytics Integrator (PAI). This session was to help you understand how cloud applications can embed predictive analytics and how data scientists and business analysts can build predictive models for cloud applications. You can watch the replay here –
I wanted to take an opportunity to summarize what PAI allows from an application perspective:
Predictive Analytics Integrator (PAi)
- SAP applications such as S/4HANA, SAP Business Integrity Screening & SAP Hybris Cloud for Customer to deliver preconfigured Predictive Scenarios embedded in their applications.
- Scores are consumed directly in the application by the decision maker.
- The lifecycle of the models is easily managed include applying, retraining and reviewing the model quality.
- Customers can adjust Predictive Scenarios delivered by SAP or create their own and seamlessly Publish to their applications from Predictive Factory.
The following are the benefits of Embedded Machine Learning in S/4HANA
- Application developers code business logic
- ML managed by Predictive Scenario container
- Stable API
- Managed model bindings
- CDS views (ABAP) or Rest Service (SCP)
End-to-end Lifecycle Management
- Full integration of ML content with Software Lifecycle Management
- PAI manages transport of Predictive Scenarios and models (trained/untrained)
- Shipment: SAP -> Customer
- on premise Dev ->Test -> Prod
- Cloud: Key User Transport
Flexibility and Agility
Customers train, run, monitor, retrain embedded predictive models
Build your own models and inject into the business process where you need it
- Custom predictive scenarios
- Bespoke models for existing predictive scenarios
- Side-by-side extensibility through PAi REST
For more details around this, please see the below: