Making Real-Time Predictions on Streaming Data
SAP Predictive Analytics 3.0 was released a few days ago and it introduces an exciting new capability to enable real-time predictions on streaming data.
The scoring equation of a classification, regression, or clustering model can be generated in an SQL-like language called Continuous Computation Language (CCL). CCL is the core language of SAP HANA Smart Data Streaming and allows you to process real-time data streams such as sensor data from IoT devices or in fact any data source you choose.
So once the model has been trained in SAP Predictive Analytics you can integrate the real-time scoring into a SAP HANA Smart Data Streaming project.
So how do you actually do this? The SAP HANA Academy has produced a step-by-step tutorial video to show how you generate a CCL script in SAP Predictive Analytics from a previously trained model and then how to integrate that script into a streaming project.
The scenario shows how an insurance company assesses past insurance frauds in order to create a category of client characteristics that may be susceptible to make fraudulent claims. The model is trained via the APL with processing delegated to SAP HANA – however any model created in SAP Predictive Analytics could be used for this scenario.
The generated CCL is added to a SAP HANA Smart Data Streaming project and tested with a real-time data stream.
In addition to regular scoring, the tutorial also shows how to make a yes/no “is a likely fraud” decision on streaming data – so that immediate action can be taken as appropriate.
Or find it here in the Smart Data Streaming playlist: SAP HANA Academy – APL: Real-Time Scoring with Smart Data Streaming
This tutorial video was created using SAP Predictive Analytics 3.0 and SAP HANA SPS 11.