Embedding Machine Learning into SAP S/4HANA

Part 6 of the blog series:

A podcast on this topic is also available here.

A blog series is in works around the topic of the ISLM framework and how it is leveraged for building machine learning scenarios!

Continuing our blog series after a brief interruption due to the unprecedented pandemic times, let us now dive into the details of how you can build machine learning and predictive analytics into SAP S/4HANA. In the earlier blog, we discussed briefly the various use cases around embedding and consuming machine learning services with SAP S/4HANA by explaining the way these use cases are organized. Let us now look into the mechanics of how this functionality is embedded into SAP S/4HANA.

In the blog series earlier, while discussing the architecture and the different approaches of doing predictive analytics and machine learning with SAP S/4HANA, we explained the concept of the Predictive Analytics Integrator (PAi) and the key role it plays while embedding this machine learning functionality. Now we are releasing the second version of the PAi which is called Intelligent Scenario Lifecycle Management (ISLM) in Q3 2020. While PAi was focused on embedding HANA’s APL library directly without coding, the HANA PAL algorithms could be embedded using generic coding methodology into SAP S4HANA. Now with ISLM technology, you could embed HANA PAL and HANA APL without any coding into the SAP S/4HANA business applications. The lifecycle management of ISLM handles this approach beautifully. Furthermore, you could also leverage the ISLM technology to leverage the other machine learning algorithms from R programming, Tensor Flow, Python libraries etc., using the side-by-side approach which we shall discuss in the next blog – “Expanding the digital core with SAP Business Technology Platform“.

Typically the need here is to handle the embedding of machine learning functionality with minimal or no movement of data. These algorithms used in embedded ML are very performance intensive as high volumes of data are handled and processed. As explained earlier with SAP HANA you have the Predictive Analysis Library (PAL) and Automated Predictive Library (APL) application libraries that provide statistical and mining algorithms. While the SAP HANA APL has built-in operations like feature engineering and detection of adequate algorithms, the SAP HANA PAL library provides more than 100 different flavors of ML algorithms. The purpose of PAi aka ISLM is to provide a common interface for the consumption of the ML models independent of the underlying predictive engine to provide predictions and results. This modeling and administrative tool is hugely helpful in managing the complete lifecycle of the model creation, model training, model adoption and model application into the SAP S/4HANA business processes.

The following pictures gives a quick overview of how ISLM can help to articulate in creating the ML models and embed into the SAP S/4HANA applications. Here we are using the PAL (also called HEMI – HANA Embedded Machine Language Interface) or APL libraries in creating the these ML models.

Step 1: Use the Intelligent Scenario Management app

Step 2: Select Embedded modeling

Step 3: Select APL or HEMI (HANA PAL)

As seen above you can start creating an embedded predictive model or use an existing predictive model into your SAP S/4HANA apps.

Currently the following is available for the customers/partners:

• All the old scenarios developed using PAi are migrated to ISLM
• New SAP S/4HANA intelligent scenarios integrated into S/4 are available based on the HANA ML (APL and PAL algorithms)
• Custom developed scenarios are possible based on embedded SAP HANA ML (APL and PAL algorithms) – these can be developed by the customers and partners

For the audience to understand the concepts of how to leverage the ISLM technology to embed machine learning into SAP S/4HANA business processes, we have come up with a best practices package. The best practices package provides a step-by-step guide on the different aspects of machine learning with SAP S/4HANA, explains the different approaches of doing machine learning with SAP S/4HANA and finally provides access to all the different use cases along with the scope items corresponding to them. We have a created 2 different scope items or documentation that explains how to use ISLM while embedding in the SAP S/4HANA Cloud as well as embedding the the SAP S/4HANA On-premise.

Additionally an SAP CAL appliance image is available for customers and partners to try out some of the example ISLM scenarios. You can find more information here about the SAP CAL appliance.

In the next blog let us discuss the mechanics behind creating an ML scenario using the side-by-side methodology.

Here are some quick links to the blogs in this series to give you a complete understanding of how Predictive Intelligence is infused into SAP S/4HANA.

• Resources and journey to machine learning with SAP S/4HANA
• Part 1 – Leveraging Predictive Intelligence with SAP S/4HANA
• Part 2 – Architecture and deep-dive of the different approaches around Predictive Intelligence
• Part 3 – Process flow leveraging Machine Learning and Predictive Analytics
• Part 4 – Scope and functionality in the context of an end-to-end process leveraging ML
• Part 5 – Activating machine learning functionality for SAP S/4HANA
• Part 6 – Building ML into the digital core of SAP S/4HANA – Embedded ML(this blog)
• Part 7 – Enhancing the digital core with ML Services (Side-by-Side ML)
• Part 8 – Extending the digital core by leveraging ML with SAP Analytics Cloud
• Part 9 – ML Extensions to SAP S/4HANA processes
• Blog series – ISLM for machine learning with S/4
• Introducing the book – Implementing Machine Learning with SAP S/4HANA

Happy predicting the future!!