Machine Learning in S/4 HANA – An Overview


Last updated on October 9, 2020


Prasad MVSR

RSS FeedRss Feed

Cost-led innovations are taking center stage as companies scale or face the constant need to disrupt themselves before the industry/market does.

Forecasting scenarios of trends, influencers, and business intelligence are now pretty much business-as-usual with the rapid adoption of technology. However, every company is at a different level of machine learning maturity – either overcrowding their enterprise systems with AI (artificial intelligence) for AI’s sake or just building sophisticated ML use cases and models. Regardless of the maturity level, modern ERP systems such as SAP consider automation, predictive analytics, conversational UI, chatbots, and innovative business process modeling as key building blocks to constitute an ’Intelligent’ enterprise. And why not?

ML algorithms are not resource-intensive in terms of memory consumption and CPI time. Plus, embedded ML architectures have a low TCO (total cost of ownership) and low TCD (total cost of development). Let us understand why the availability of big data, better algorithms and processing power facilitate easy and quick implementation of machine learning into SAP S/4 HANA’s architecture.

Figure 1 – This diagram depicts embedded ML architecture based on HANA ML with PAL (Predictive Analytics Library) and APL (Automated Predictive Library) the necessary algorithms.

A trusted path to an intelligent enterprise

There are countless ways that ML (machine learning) driven predictive intelligence and automated workflows benefit a business today – across any industry. You can use predictive modeling to assess your customer churn or apply NLP (natural language processing) and OCR (optical character recognition) to cut downtime in your packaging process. Marketers are used to seeing live dashboards of customer sentiments via social channels, while bankers are improving customer experience by resolving disputes quickly with RPA (robotic process automation).

In the case of S/4HANA’s in-memory platform, the power of SAP’s intelligent suite and digital core combines the transactional and analytical capabilities to innovate freely with embedded ML abilities. Additional Leonardo ML features that are pre-integrated into S/4HANA support businesses by deploying specific use cases and out-of-the-box solutions within the HANA suite.

Another benefit of S/4HANA ML models is that it can easily handle large volumes of data and GPU. Typically exchanged data files such as images, audio, video, etc. require big data solutions. Scenarios based on neural networks, sentiment analysis, image recognition, and NLP also come with high data volume and demand GPU power. S/4HANA enables these capabilities through SAP CAI or Conversational AI – a self-learning chatbot building platform that uses ML functionalities to gain knowledge based on historical experience and data. Similarly, Fiori helps with rich UI interfaces, BI visualization, and charts with ML.

In terms of use-cases based ML scenarios, businesses can leverage HANA’s model adaptation and configuration techniques to meet both ‘moderate’ and ‘complex’ AI requirements for their business. SAP’s ABAP CDS links your apps to the ML data seamlessly to increase ‘predictive confidence’ and ‘predictive power’ and validate your ML models.

Accelerate your last-mile to ML deployment

As the illustration above shows, SAP HANA does facilitate intelligent applications for every industry across every continent. Many companies, however, struggle and waste nearly 25% of their time in deployment.

As an SAP Gold Partner for 25+ years, YASH specializes in capturing the opportunities between the misaligned pace of work, processes, and machine learning teams. We help turn your data mine into a gold mine (so to speak) of valuable opportunities and performance gains.
Read further about our SAP HANA ML success stories and offerings here.



No Comments

Add Comments

Type in a topic service or offering and then hit enter to search

Thank you for your message. It has been sent.