A podcast on this topic is also available here. Let us now continue into our series and understand the details behind consuming the machine learning services from SAP Cloud Platform. In the earlier blog, we discussed briefly the various use cases around embedding and consuming machine learning models with SAP S/4HANA by explaining the way these use cases are organized. Now we will look into the mechanics of how this functionality is consumed by SAP S/4HANA.
In the blog series earlier while explaining the architecture and the different approaches of doing predictive analytics and machine learning with SAP S/4HANA, we discussed the concepts behind how SAP Leonardo foundation is leveraged in building these machine learning services. Now that SAP Leonardo is merged into SAP Data Intelligence that runs on the SAP Cloud Platform, we shall look into how the machine learning models are built to be leveraged by SAP S/4HANA – we also call this approach the side-by-side ML. While the embedded ML targets on business logic and machine learning algorithms residing in SAP S/4HANA, the side-by-side ML targets machine learning algorithms residing on SAP Cloud Platform (typically in SAP Data Intelligence or the SAP AI foundation) and the business logic stays either on SCP or SAP S/4HANA depending on the application requirements.
The side-by-side ML scenarios are scaled out and built on the SAP Cloud Platform to ensure the load on SAP S/4HANA systems is low and the runtime of the systems is acceptable. With the new concept of the ISLM technology, you could leverage the SAP HANA ML algorithms or non-SAP ML algorithms from the other libraries such as R programming, Tensor Flow, Sci-Kit Learn or Python etc. The ML services could be built using a mix of these machine learning libraries and made available for consumption by the SAP S/4HANA applications or SAP Cloud Platform applications. The SAP S/4HANA extension applications consume SAP Data Intelligence capabilities with the business data on SAP Cloud Platform (irrespective of where the application lives, either SAP S/4HANA or SAP Cloud Platform) and ML algorithms from SAP Cloud Platform following the golden rule of bringing the algorithms to the data!
SAP Data Intelligence is an important component of the side-by-side ML scenarios which is designed for cloud, hybrid or on-premise landscapes. The data scientists can use the machine learning functionality of SAP Data Intelligence with the different tool sets to design, create and train the models along with managing the life cycle of the ML models.
There are 2 different options of leveraging the side-by-side ML models on SAP Cloud Platform. While one option is to re-use the AI business services that are already built on the SAP Cloud Platform as shown in the figure below,
the second option is to create new ML services using SAP Data Intelligence. The following briefly explains the process using SAP Data Intelligence to create ML models.
Step1: Launch SAP Data Intelligence
Step 2: Create an ML model with SAP Data Intelligence
Step 3: Add input datasets etc., for creating and training the model
The final results and the output of the predictions are consumed into the SAP S/4HANA business applications or the SCP applications. In an earlier blog, we have explained how the ML service is configured with the communication scenarios to be integrated and consumed into the SAP S/4HANA business processes.
In the next blog let us discuss the mechanics behind creating an ML scenario using the explorative analytics methodology using SAP Analytics Cloud.
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.
Happy predicting the future!!