SAP Best Practices for SAP Integrated Business Planning: Artificial intelligence (AI)

[InaGlaes]

In the last blogs, you learned a lot about the benefits, the different assets, the sample planning areas, the unified planning process, integration and so on of SAP Best Practices for SAP Integrated Business Planning for Supply Chain (SAP IBP). Let's conclude this blog series with SAP IBP Artifical Intelligence (AI) and the embedded and side-by-side intelligent use cases, SAP IBP AI offers. It doesn’t need an extra explanation, that machine learning and AI can make your life easier and help you to optimize your supply chain planning, to drive planning accuracy and automation. With that, you can free up the capacity of planners to take care of critical topics instead of daily routines, you can lower safety stocks, improve your master data quality, improve forecast quality, and much more.

Let me also state, that this is just the beginning. You will see more use cases in the next releases, for example helping you to find your way in the system, understand the results of planning runs better, get intelligent proposals for critical situations, and much more.

There is a quite long list of already existing AI use cases, let me focus on a few with the related Best Practices solution processes – just to make you curious.

Master Data Anomaly Detection

To have correct and consistent master data is for sure one of the biggest challenges in IT. An easy way to improve the master data quality is to run a master data consistency job after the master data has been imported to SAP IBP. Machine learning can learn semantic rules to identify problems in master data and recommend values for correction. In the solution processes IBP – order-based planning inbound integration with SAP S/4HANA (2S4) and IBP – order-based planning inbound integration for deployment planning (4VA) we run the ML Master data Consistency job on attributes PRODGROUP and PRODTYPE of the data type PRODUCT. This is just one example and you can enhance it to your needs.

Automated cluster and threshold proposal in segmentation

ABC/XYZ segmentation is a method of grouping planning objects based on key figure values. There are two types of segmentation: 

• ABC segmentation is the prioritization of planning objects based on their relative importance. In the SAP Best Practices solution process, the product combinations are categorized based on their revenue. For example, the products contributing to 70% of the revenue are classified as A-products. 
• XYZ segmentation is the classification of planning objects based on their demand volatility. In the SAP Best Practices, the product combinations are categorized based on their actual quantity. For example, products with a rather constant demand are classified as X-products. 

Why is segmentation important? Segmentation helps you to generate more accurate results for demand planning, inventory planning, and sales and operations planning.

SAP IBP offers different methods for segmentation, and one of them is K-means. If you choose this method, the system uses machine learning to create segments as homogenous as possible. This is especially useful if you are not sure what thresholds should be defined for the segments.

Here is the link to the related solution process IBP – ABC-XYZ segmentation (1S6).

Change point detection

Time-series analysis and change point detection allow you to identify patterns in individual time series. This enables you to learn about major changes in the sales history and allows you to select the most suitable forecast algorithms.  

Time-series analysis is used to detect patterns in a time series. The most important patterns are Trend, Seasonality, Continuous, Irregular, and combinations thereof. The patterns are saved to an attribute called Time Series Property. 

Change point detection allows you to identify significant changes in a time series. There can be level shifts (when the mean of the time series values alters significantly) or trend changes (when the direction or slope of a trend alters significantly). 

Here is the link to the related solution process IBP for demand – time-series analysis (34Y).

Forecasting

In the solution process IBP for demand – demand planning (11V), we provide different forecast models, for example for A- und B- products, for C-products, and products with intermittent demand, with automatic outlier correction, and so on. One of them uses the following advanced forecasting algorithms:

• Gradient Boosting of decision trees
• Extreme Gradient Boosting 
• ARIMAX

 We won't explain here in this blog the details of the different algorithms, you can use the links to the help portal if you need further information.

Demand Sensing

Similar to the gradient boosting of decision trees algorithm, demand sensing with gradient boosting is an ensemble machine learning algorithm that considers the information gained from various external signals when the sensed demand is calculated. It’s used in the solution process IBP for demand – demand sensing (11X).

Let me point you at the end to the blog of my colleague Laura Tozzo - Unlock your supply chain potential with SAP Integrated Business Planning AI - to get more examples of how artificial intelligence capabilities in SAP IBP significantly drive planning accuracy and automation.

Was this blog useful for you? You can continue your SAP Best Practices for SAP Integrated Business Planning for Supply Chain learning here.

And here you can find some general links: 

• Back to the overview blog
• Process Navigator 
• Process Navigator direct link to SAP Best Practices for SAP Integrated Business Planning (latest ver...