SAP Machine Learning: Wheres the Beef?
Level 1 – Easy ; 30 minute read
Audience: Project managers, business analysts, subject matter experts
Author: Mark Muir, SAP BTS, S/4HANA RIG Americas
Welcome! Where’s the beef is an 80’s catch phrase (fast food advertisement) that is synonymous today with lack of substance or detail. The what and where of machine learning to date has been easy to comprehend but the how (e.g. how can algorithms learn from data without having to be explicitly programmed or what technologies support machine learning?) has been intangible. The details behind algorithms will be left for another day, the realization behind technologies that can irreversibly alter the fabric of economies and 1industry strategies is worth looking into further.
Under pinning SAP’s strategy helping customers increase productivity by enabling the workforce to focus on higher value tasks and driving down repetitive tasks through automation is the Intelligent Enterprise. Comprised of 3 components the Intelligent Enterprise is uniquely positioned to deliver a framework that shares end to end integration, industry expertise and embedded intelligence.
- Intelligent Suite– Unified process landscape
- Digital Platform– Intelligence delivered at scale in the cloud
- Intelligent Technologies– Embedded technologies to drive business outcomes
Fig. 1 – Vision of the SAP Intelligent Enterprise
1. Intelligent Suite
The Intelligent Suite delivers intelligence across the value chain with intelligent applications for every line of business, this includes industries (25) and countries (180+) all supported by S/4HANA. Each line of business delivers extended capability and outcome by application as shown (fig. 1). A formidable line of business portfolio delivers next practices to our customers.
- Customer Experience e.g. SAP C/4HANA
- SAP Integrated Business Planning e.g. SAP Integrated Business Planning
- Manufacturing & Suppy Chain e.g. SAP Digital Manufacturing ; SAP Digital Supply Chain
- Digital Core e.g. SAP S/4HANA Cloud
- People Enablement e.g. SAP SuccessFactors
- Network and Spend Management e.g. SAP Fieldglass, SAP Ariba and SAP Concur
Fig. 2 – Intelligence Suite: Intelligent across the Value Chain
2. Digital Platform
Aging enterprise landscapes are under pressure; business, devices, social media is generating an overwhelming amount of data (structured and unstructured) creating unique challenges and new opportunities.
- How to manage data from any source, in any format and rapidly develop, integrate, and extend business applications with an open digital platform?
- How to improve productivity, innovate while still running today’s business?
- How to get actionable insight from external data sources to respond faster to customer needs and competitive threats?
- Satisfy demand for a more purposeful work environment with new workforce skills?
As a core component of the Intelligent Enterprise Digital Platforms modernizes back-end systems and delivers to the enterprise data-driven intelligence and innovation.
Fig. 3 – Digital Platform: Unlock data-driven intelligence and innovation
Digital Platforms for the Intelligent Enterprise
The SAP Digital Platform is not one single platform it is a combination of platforms with many supporting applications and capabilities more information here.
- Application Integration and Infrastructure
- Delivering deep data and process integrations through APIs and micro-services more
- Real-time data at your fingertips with SAP HANA more
- Enterprise Information Management
- SAP HANA powers SAP applications as the foundation of high-performance data warehousing and analytics more
- SAP HANA
- Platform for extending the business processes of the Intelligent Suite and enabling new innovations more
- SAP HANA Data Management
- SAP Data Hub provides data orchestration and metadata management across heterogeneous data (fig. 4) more
- SAP Cloud Platform
- Connect to and integrate with enterprise systems (e.g. via CP connector or SAP DataHub), e.g. S/4HANA or SAP HANA DB (fig. 5, 6) here
Fig. 4 – SAP Machine Learning Foundation & Data Hub Connectivity
- Combining the management, governance and availability of data across platforms and systems helps generate innovative services on the SAP Cloud Platform.
Fig. 5 – Combining Data Management and Cloud Platform
- SAP Cloud Platform also works together with HANA Data Management Suite (HDMS) in many ways. (fig.6)
- Integrate with HMDS and other backend systems connecting to SAP back end, non-SAP system and data source
- Leverage HANA services for data insight
- Build a great user experience
- Leverage SAP Cloud Platform development tools to build custom apps.
- Leverage SAP Leonardo and implement IoT solutions using SAP Cloud Platform
- Use SAP Cloud Platform to manage Big Data as a Service
Fig. 6 – Combining Data Management and Cloud Platform
3. Intelligent Technologies
Intelligent Technologies depend on large amounts of enterprise data generated by core applications residing within various systems and many customers use different solutions that focus on providing analytical results based on this data. With intelligent applications embedded into core applications, business processes now have the ability to smartly utilize enterprise data as training data for the implementation of Machine Learning services. Businesses are learning and evolving from historical data and training models are helping drive down the use of repetitive tasks through automation (Fig. 7, 8).
Fig. 7 – SAP Leonardo: Intelligent Technologies
Helping customers realize the value of Machine Learning, SAP has introduced Industry Innovation Kits which provide a service combining design-thinking and industry accelerators to help derive value from innovative technologies quickly and with reduced risk. All available by industry profile i.e. Discrete, Consumer, Service and Energy & Natural Resources read more …
As you can imagine the individual technologies and applications supporting Machine Learning is extensive, below are main application frameworks.
- (a) SAP Machine Learning Foundation
- (b) SAP Cloud Platform
- (c) SAP Predictive Analytics
- (d) SAP Lumira
Fig. 8 – SAP ML Foundation: Enabling customers/partners to use and train their own models
(a) The SAP Leonardo Machine Learning Foundation is aimed at solving the issue of how by providing an enterprise-grade platform for Machine Learning in the cloud by unlocking valuable insights from structured and unstructured data using machine learning technology. With the help of easy-to-use APIs, you can use the foundation to enable intelligent enterprise applications, key benefits here (see fig. 9).
Fig. 9 – SAP Machine Learning Foundation
The SAP Machine Learning Foundation is designed around 3 architecture tiers (Fig. 10)
- Application Tier
- SAP Leonardo ML Foundation API‘s
- Activity: Inference apps
- Compute Tier
- TensorFlow Serving Model Containers
- Activity: Deploy model for inference
- TensorFlow Training Containers
- Activity: Used only during training
- Persistence Tier
- Model Repository
- Activity: Upload model
Fig. 10 – SAP Machine Learning Foundation – Architecture Tiers
- Key Message: SAP ML Foundation is a toolbox of intelligent technologies, micro-services and data management tools (core components) residing on the SAP Cloud Platform that are the basic building blocks required to quickly and easily enrich applications (bring or deploy your own model, create new machine learning services i.e. out of the box and co-innovation).
- Technology: Machine learning technology can be quickly incorporated into existing SAP or non-SAP solutions using the Machine Learning services provided by SAP Leonardo Machine Learning.
- Functional Services: Exploit a variety of data types covering image, speech, text and time-series data. E.g. detect objects in images, detect keywords in text or predict time series data.
- Core Capabilities: Customize ML services with own data or deploy custom models and consume them via REST APIs
- Users: Developers (quickly and easily infuse apps with ML without data science know how) and data scientists (cater towards enterprise-grade operation of ML and lifecycle management)
- Application: End-user apps can be infused with intelligent ML services
- Enterprise System: Connect to and integrate with enterprise systems (e.g. via CP connector or SAP DataHub), e.g. S/4HANA or SAP HANA DB.
(b) The SAP Cloud Platform (enterprise PaaS) is an open platform as a service for developing and running business applications in a fully provisioned environment e.g. SAP ML Foundation. Using a set of end-to-end services, capabilities, APIs, and tools, you can build new cloud applications as well as extend and integrate applications in the cloud. You can also use SAP Cloud Platform to extend and integrate with existing on-premise systems (hybrid scenarios) or other cloud products. (ref. featured scope description). (fig. 11)
Fig. 11 – SAP Cloud Platform Highlights
- quickly extend and personalize existing cloud and on-premise apps regardless of vendor applications – with choice in cloud providers, open and common development languages
- optimize for maximum gain by connecting your cloud and on-premise applications and integrating your entire IT landscape and business processes
- innovate rapidly on a cloud platform with machine learning, big data, Internet of Things technologies simplify building applications to solve existing or new business problems.
- See Additional Information to learn more
SAP Cloud Platform architectures
- SAP Fiori Cloud
- Extend SAP Cloud Applications
- Internet of Things Innovation
- Cloud Foundry – Build new apps and micro-services
- SAP Cloud Platform API Management (Fig. 12 also read SAP ML Foundation introduction)
Fig. 12 – SAP Cloud Platform API Management Architecture – High Level
(c) SAP Predictive Analytics is a statistical analysis and data mining solution that enables you to build predictive models to discover hidden insights and relationships in your data, from which you can make predictions about future events. Built for both data scientists and business / data analysts, it makes predictive analytics accessible to a broad spectrum of users. (fig. 13)
- Used to automate data preparation, predictive modeling, and deployment tasks
- PAL, BFL and R language support
- Advanced visualization
- Native integration with SAP HANA
- S/4HANA Backend Machine Learning Technologies
- S/4HANA Embedded Predictive & PAI the Framework
Fig. 13 – Embedded Predictive Analytics Architecture
(d) SAP Lumira enables customers to gain insights from trusted enterprise data sources and personal data, and to share those insights through interactive visualizations. The application has connectivity to SAP enterprise data models in SAP BW, SAP HANA and SAP BusinessObjects universes (UNX) as well as to a wide variety of third party databases link. (fig.14)
- Numerous predictive and visualization charts available with Lumira.
- Statistical summary chart
- Box plot
- Parallel co-ordinate plot
- Bubble plot
- Scatter plot matrix
Fig. 14 – SAP Lumira Visualization
I hope you enjoyed this blog and are able to articulate how the Intelligent Enterprise supports AI/ML in a data-driven economy.
SAP Technical l What’s New
- SAP Cloud Platform – What’s New+All Release Notes
- To get notifications of published release notes, please subscribe at the SAP Community WIKI
SAP S/4HANA Discovery
- SAP Best Practices Explorer – S/4HANA Enterprise Cloud , On premise
- SAP Roadmap Viewer – Cloud , On premise, Cloud – Single Tenant Edition
- SAP help.sap.com – S/4HANA 1909 , S/4HANA Cloud
- SAP Product Availability Matrix (PAM) – S/4HANA
SAP Cloud Platform
- sap.com – SAP Cloud Platform
- What is SAP Cloud Platform
- SAP Roadmap / SAP Cloud Platform
- SAP Cloud Platform Public wiki
- SAP Community Network / SAP Cloud Platform
- SAP Cloud Platform – Free Trial
SAP Predictive Analytics
SAP Leonardo Machine Learning Foundation
SAP Knowledge Management
- SAP Machine Learning – Learning Journeys: Developers , Data Scientists
- openSAP – SCP courses , Enterprise ML in a Nutshell , Deep Learning with Tensor Flow , Freedom of Data with SAP Data Hub , SAP Leonardo Machine Learning Foundation
Further reading in this Machine Learning blog series