Skip to Content
Author's profile photo Sainath Kumar

SAP Leonardo Machine Learning – Overview


There has been a lot of buzz around SAP Leonardo, which includes a portfolio of offerings ranging from IoT, Blockchain, Machine learning, Big Data, Data Intelligence,Design Thinking & Analytics. This blog focuses on the SAP Leonardo Machine Learning and provides the readers with an understanding and overview of the current features offered.

For an SAP customer, the following questions will arise –

  1. “What is SAP Leonardo Machine Learning?”   
  2. “Is it a new machine learning product from SAP?“
  3. “What happens to existing SAP Advance Analytics tools (SAP Predictive Analytics , HANA PAL (Predictive Analytics Libraries) ?”
  4. “How will it integrate with existing SAP applications?”



Let’s begin with answering the first question, SAP Leonardo Machine learning is not a new product offered by SAP. SAP Leonardo Machine Learning is the new branding for SAP’s entire portfolio of intelligent applications and services. It ranges from well-established product offerings, such as SAP HANA Predictive Analytics Library(PAL), SAP Predictive Analytics(SAP PA 3.x) to services offered through the SAP Leonardo Machine learning foundation. The services offered are explained below.


The below diagram gives an overview of SAP Leonardo Machine Learning components.

Breaking down components of SAP Leonardo Machine Learning

Let’s try to understand how SAP is trying to position the portfolio of Machine Learning Applications. This portfolio includes the following three components

  1. HANA(PAL)- Which is part of HANA’s application function library (AFL) defines functions that can be called from within SAP HANA SQLScript procedures to perform predictive analytic algorithms
  2. SAP Predictive Analytics – Standalone software application provided which can be used in conjunction with HANA(PAL) to build predictive models .
  3. SAP Leonardo Machine Learning Foundation on SAP cloud platform consists of machine learning API’s available via API Business Hub which can be implemented as REST APIs. (A RESTful API is an application program interface (API) that uses HTTP requests to GET, PUT, POST and DELETE data.

The following diagram describes the different services provided by the machine learning foundation.

SAP Machine Learning Functional Services –Provide readily consumable pre-trained models that can be used as a web service by calling simple REST APIs. SAP now has a growing list of consumable APIs on the business hub. Here are some of the examples of these APIs:

  • Document Feature Extraction API
  • Image Classification API
  • Image Feature Extraction API
  • Time Series Forecast API
  • Topic Detection API
  • Product Text Classification API
  • Similarity scoring API

SAP Machine Learning Predictive Services – SAP offers predictive services which can perform analytics on data on SAP HANA DB on SAP Cloud platform. Some of the services offered are listed below:

  • SAP Predictive Analytics Integrator Service*
  • Clustering service
  • Dataset service
  • Forecast service
  • Outlier service
  • Recommendation service
  • Whatif service

SAP Predictive Analytics Integrator Service -This service integrates and consumes predictive models within cloud applications and enables productive utilization of the predictive models. SAP Predictive Analytics Integrator is a framework to enable SAP Applications to embed predictive models natively in enterprise applications. Business users can use the output of the models in the form of insights, scores and decisions to help them in their day to day work while data scientists can debrief the models to understand how they work and update them to increase accuracy and ROI.

SAP Leonardo Machine Learning Business Service – The services provided by SAP focus on business specific use cases and out of box solutions. For example, SAP Leonardo Machine Learning foundation can enable service organizations, by easily categorizing and smartly processing incoming service inquiries, or by analyzing historical activities of business network users. Some of the services called as intelligent services are listed below:

  • SAP Intelligent Financing API
  • SAP Service Ticket Intelligence Classification
  • SAP Service Ticket Intelligence Recommendation

To understand how SAP is positioning these APIs, we can draw parallels from Bluemix platform which offers Watson APIs (Text to Speech API, Personality Insights API) or AWS platform offering Machine Learning APIs. The blog does not provide a comparison between the API’s offered by SAP and other API providers. The blog portrays SAP’s vision and future road map for providing machine learning APIs within its ecosystem . Third party providers also can host their APIs on this SAP cloud platform.

Technical Scenarios

Let’s explore various technical scenarios to understand how customers will leverage SAP Leonardo Machine Learning Platform. We go with the assumption that the customers understand the need to set up a next generation platform for predictive and cognitive analytics. Many customers have already realized the importance of having an enterprise wide adoption of machine learning as compared to having niche applications run by group of data scientists.

  • Customers running HANA (PAL) – HANA will soon provide an Application Function Library (AFL) to natively access and consume machine learning services provided by SAP Leonardo Machine Learning Foundation. This will enhance HANA’s already existing PAL algorithms with service APIs for certain applications.
  • Customers running SAP Predictive Analytics –SAP Predictive Analytics Integrator works with on premise and cloud based applications and is used to solve business problems based on structured relational data. Customers now can build models in SAP PA and consume them in real business applications
  • Customers running open source machine learning applications (R, Python, SparkML..) SAP Leonardo Machine Learning services can be easily implemented as REST APIs

Watch this space for more clarity from SAP

  • SAP has recently introduced a set of following cognitive APIs – Image processing, Topic detection, Product Text Classification Et al., customers need to evaluate and compare the functionalities and features (performance, security, compatibility, scalability, pricing ..) of each API with features of similar APIs provided by leading providers.


The diagram below combines all the components of SAP Leonardo Machine Learning (We do not have an exhaustive list of all APIs)

All API are available @

Assigned Tags

      You must be Logged on to comment or reply to a post.
      Author's profile photo Antoine CHABERT
      Antoine CHABERT

      Nice recap! Just want to add that the SAP Predictive Analytics team is also in charge of a SAP HANA called APL (for Automated Predictive Library) that makes it possible to create Automated models either using scripting capabilities of SAP HANA or using the delegation capabilities of SAP Predictive Analytics.

      Author's profile photo Pavan Golesar
      Pavan Golesar


      What a wonderful explanation of building blocks of SAP Leonardo ML Foundation.  Keep it up! 🙂


      Pavan Golesar

      Author's profile photo Mohammed zahid
      Mohammed zahid

      Dear @Sainath Kumar , and @Abdel DADOUCHE


      Nice blog and gives good overview on ML .

      I have few questions on ML could you help me to answer

      1. I have build a PoC on SAP ML foundation (here ) , for the real time implementation , is there any methodlogy we follow ? like Agile /ASAP/Waterfall / or SAP Cloud implentation goes same way the other projects
      2. If we want to do Real ML implemenation like SAP HANA (PAL) or Predictive Analysis ? How much Machine learning knowledge is required ? I mean orginal machine learning (Regression/classification/ cluster/ etc ) or Python knowledge would be sufficient ? As what I learned so far is we import libraries/ spliting the dataset /scaling / deploy ML regression from Sklearn /plot it etc..  So I want to understand where the real machine learning comes into the picture
      3. When we say SAP HANA (PAL ) we have good blogs like here   ( which is in detail and excellent ) will be sufficient to follow ??

      It would be great if you guys can  clear the vision of SAP Leonarado Machine learning for me .

      Either SAP HANA(PAL)/ Predictive analysis or CLoud foundry is good to go ahead.


      Thanks and Regards,


      Author's profile photo Abdel DADOUCHE (SAP)
      Abdel DADOUCHE (SAP)

      Hi Mohammed zahid


      There would be so much to say about the few point that you highlighted in your previous comment.

      The blog series I started last year was meant to provide a brief introduction about some of the fundamentals that anyone should know before getting started like the algorithm learning styles or the project methodology. This is probably a good first step.

      The openSAP course Getting Started with Data Science is going a step further with more details but that's not the end there too.

      There are so many techniques and algorithms out there that it's almost impossible to know them all. We just need to keep in mind that each algorithms and techniques have strength and weaknesses.

      SAP Leonardo ML Foundation provides access to:

      • a set of REST API that allows developers to easily integrate the use of pre-built models into their apps without any ML/Data science knowledge
      • a set of customizable pre-trained models that can be partially retrained with customer datasets to address customer needed without an extensive ML/Data science knowledge
      • a platform that allows you to either Bring Your Own Model or Train Your Own Model if you have ML/Data science knowledge

      SAP Leonardo ML Foundation was built to consume "unstructured" content like image, text or audio/video and be used by apps deployed on SAP Cloud Platform like S/4HANA Cloud or an SAP SaaS solution extensions.

      SAP HANA with its APL & PAL libraries provides access to algorithms (not pre-trained models) and can be used as a "platform" (or environment) to Train Your Own Model using data data stored into a structured format (a table or a view for instance).

      Keep also in mind that SAP HANA is the foundation of solutions like S4/HANA, Hybris Marketing and many other that already provide (or OEM) predictive capabilities.

      Also something really nice that was recently released are the SAP HANA API for Python and R. This means that if you learned Python or R, you can now get access the SAP HANA libraries of algorithms too.

      You can check Arun Godwin Patel blog series about the SAP HANA Python library:




      Once again, I don't mean here to provide a full answer, but I just wanted to share with you what might some next steps keeping in mind that this is a learning journey that was driven for me by curiosity and enjoying sharing with others.

      Author's profile photo Vivek Chaudhary
      Vivek Chaudhary