For any questions please feel free to contact us.
Machine Learning: Predicting Delivery Date Delays in SAP S/4HANA with Embedded AI (Part 4)
(4 min read)
In the fourth part of our blog series on the Intelligent Enterprise, we explain how you can integrate machine learning (ML, a subfield of artificial intelligence/AI) into SAP S/4HANA processes, which preconfigured ML scenarios SAP is already delivering in the areas of purchasing, sales and supply chain, and what benefits machine learning brings to your business processes.
Machine Learning in the SAP Product Portfolio: An Overview
The design and operation of ML models and scenarios implemented in various SAP solutions follows the cross-industry standard process for data mining. The process starts with understanding the business requirement, followed by the steps of “understanding and preparing the relevant data”. This is followed by modeling, training, evaluation, deployment, and monitoring of the respective ML model.
Common ML algorithms are used for this purpose. For moderate ML requirements in SAP S/4HANA like forecasting, trending and influencers, the algorithms like regression, clustering, classification and time series can be used. These algorithms are part of the HANA database or are included in the scope of delivery of many SAP products, e.g. SAP S/4HANA or SAP SuccessFactors, keyword here: “Application Embedded AI”.
More complex ML requests with external data and more complex algorithms and requirements such as image recognition, sentiment analysis, natural language processing and deep learning are typically offered via a side-by-side scenario through the SAP Business Technology Platform (BTP) and are grouped under the “AI Business Services” product portfolio. AI Business Services are provided as API function calls and can be integrated into any SAP and non-SAP systems.
Furthermore, SAP plans to offer an AI Foundation on the BTP in Q3 2021, which will further simplify the management, training, and deployment of ML models (including those from external vendors) across SAP solutions.
Preconfigured Machine Learning Scenarios in SAP S/4HANA
SAP has already delivered about 20 ML scenarios with the SAP S/4HANA 2020 release and 24 ML scenarios with SAP S/4HANA Cloud 2105. Here is an overview of the scenarios from the areas of purchasing, supply chain and sales (excerpt):
|Line of Business||Scenario|
|Purchasing||Prediction of delivery date for purchase order items|
|Purchasing||Prediction of cash discount losses for invoices with payment blocking|
|Purchasing||Prediction of fulfillment rate of purchase contracts|
|Supply Chain||Prediction of delivery delay for stock transfer order|
|Supply Chain||Early detection of slow / non-moving stocks|
|Supply Chain||Prediction of stock transfer product lead time|
|Sales||Prediction of schedule delays for outbound shipments|
|Sales||Prediction of quotation conversion probability rate|
|Sales||Sales performance prediction|
The ML models are fully integrated into the processes and generate forecasts based on historical process data. The forecast results are overlaid on the operational and analytical apps, enabling employees to react to specific situations much earlier or proactively.
Let’s take a closer look at the scenario “Predicting schedule delays for outbound deliveries”.
Predict Schedule Delays for Outbound Deliveries in SAP S/4HANA
An inside sales representative can oversee the creation and processing of the scheduled delivery to the customer as part of the order fulfillment process. The sales employee is also able to identify problems immediately and initiate appropriate measures. This ranges from the timely supply of the procurement processes to the creation of deliveries as follow-up orders to transport planning, picking, packing, and shipping in the delivery process and transport processes.
In the SAP Fiori app “Predicted Delivery Delay”, a delay in days for delivery creation and processing as well as the predicted overall status of the delivery item is displayed for each sales order item in addition to the planned delivery date. In addition, the statistics for all order items are displayed graphically.
This enables sales staff to immediately identify delays in the supply chain, take timely action in critical positions and thus increase customer satisfaction.
The machine learning model on which this scenario is based currently contains around 130 standard data fields or input parameters that can be used for training. When training the model, the data field catalog can be restricted or filtered.
Are You Curious for More?
Then read on! You can find more on the topic of “Intelligent technologies in the SAP S/4HANA context” in our Intelligent Enterprise blog series:
- Part 1: Your Path to the Intelligent Enterprise: Intelligent Technologies in SAP S/4HANA
- Part 2: Intelligent Technologies as Integral Part of SAP S/4HANA
- Part 3: Conversational AI: Chatting and Speaking with the SAP S/4HANA System
- Part 4: Machine Learning: Predicting Delivery Date Delays in SAP S/4HANA with Embedded AI
- Part 5: Software Robots Relieve Your Customer Inquiry Management – How RPA Works