Technology Blogs by SAP
Learn how to extend and personalize SAP applications. Follow the SAP technology blog for insights into SAP BTP, ABAP, SAP Analytics Cloud, SAP HANA, and more.
Showing results for 
Search instead for 
Did you mean: 
0 Kudos
The latest release of smart data streaming has added some new machine learning functionality for both SAP HANA studio and the streaming runtime tool for SAP Web IDE.

Support for Machine Learning in the Streaming Runtime Tool

In previous releases, machine learning models could only be used within HANA studio. Now, you can also create and manage these models through the streaming runtime tool, where they’re called “PAL models”.

Working with machine learning in the streaming runtime tool is easy: within a PAL Models folder, either right-click a data service folder to add a model, or right-click a model to edit, delete, reset, or rename it.

Above: Adding a model to a data service in streaming runtime tool.

Above: Editing a model in streaming runtime tool.


For steps on creating machine learning models through the streaming runtime tool and inserting them into a project, look over the workflow here.

Importing Models, and a New Function: Decision Tree Scoring

We’ve also introduced a new machine learning function for smart data streaming: decision tree scoring. This function relies on data trained using a decision tree training model, which you need to import from SAP HANA.

Let’s break that down a bit. A decision tree helps determine an appropriate action or decision from a pre-determined set for a given situation. Using a decision tree model can help you identify which factors you need to consider in your decision, and how they may have influenced different outcomes in the past.

Decision tree models need a set of classified samples for training. You can’t train a decision tree directly in smart data streaming – and that’s where importing comes in.

Once you import a trained decision tree model from SAP HANA, you can create a decision tree scoring model in smart data streaming, then score and use the model within your streaming projects.

To import a model, you need to give it a name, a description, and select the table and schema to use.

Above: Importing a model into streaming through HANA studio.

Above: Importing a model into streaming through the streaming runtime tool.


It’s as simple as that – your imported model will then be ready to use with a scoring model in your streaming projects.

For more information on importing a model, see the topics here and here.

For more information on decision tree training models, look over the HANA machine learning documentation here.

Be sure to also see the topics in the section here for information on all aspects of machine learning model management in smart data streaming.