Part 5 of the blog series:
As of this writing in the last week of January 2020, we have about 35 to 40 use cases that have been built around the different lines of business and industries leveraging the Machine Learning and Predictive Analytics algorithms. In due course the number of ML use cases will be changing since some of the use cases shall be deprecated and a few of them will be updated and also you will find some new use cases based on customer feedback and customer requirements. Connecting back to my earlier blog about the different approaches that could be leveraged for infusing intelligence into SAP S/4HANA, let us now review in detail how to realize the functionality.
Though you find complexity in SAP technology or SAP software, you will understand that there is a structured approach to dissect the information and understand how to access, implement and extend these functionalities. The beauty is encapsulated in the 3 letter acronym of the scope items. Any functionality can be activated or de-activated by choosing the corresponding scope item (3 letter scope item). You will also notice that some of these scope items would have pre-requisite scope items that have to be activated and implemented before you continue further.
Let us now dive into the scenarios and understand the mechanics behind the build and implementation! Here are the different approaches starting from “embedding predictive models in SAP S/4HANA”, followed by “consuming ML services on the SAP Cloud Platform”. We shall discuss “leveraging the predictive services from SAP Analytics Cloud” in a later blog while talking about extending the digital core with SAP Analytics Cloud Predictive services.
Embedded Predictive Scenarios:
In the context of embedding predictive models into the SAP S/4HANA business processes, there are a few steps that have to be followed. Here are the steps to be followed.
System access – The system is accessible via the Fiori LaunchPad. The system administrator provides the URL to access accordingly, the various apps assigned to your role.
Roles – The “Analytics_Specialist” role is needed to first create the predictive model version, then train the model and finally activate the model.
Preliminary steps – Creation of business data for the specified scope items and any pre-requisite scope items there-of!
Select and train the model based on the data set provided or applied.
Set a model version to active that will be used in the embedded application.
Change Role – Login as the specific end user to access the app and run the functionality to see the predictions.
Now let us see an example of how a particular predictive functionality is enabled, applied and run in the context of a technical scope item. Eg., Quantity Contract Consumption – scope item 1QR.
a) Scope item and flow:
1QR – The purchaser can analyze a high-level overview of important information, such as expiring contracts, overdue purchase orders, or urgent purchase requisitions, as well as an overview of different procurementKPIs. That information can then be used to predict full consumption of a contract based on factors such as a historical data, other available influencing parameters, and so on.
b) Roles:
All of the following roles should to be assigned to be able to work with the Quantity Contract Consumption KPI.
Business Role Name | Business Role ID | Log On |
|---|---|---|
Analytics Specialist | SAP_BR_ANALYTICS_SPECIALIST | Please ask your system administrator to assign the roles to the testers. |
Buyer | SAP_BR_BUYER | Please ask your system administrator to assign the roles to the testers. |
The predictive model training needs to be done by an analytics specialist.
The analytics specialist requires the business catalog SAP_BW_BC_UMM_PC.
To use the Quantity Contract Consumption app, the business catalog SAP_MM_BC_PUR_STRATEGY must have been assigned to end user (this business catalog is also included in the business roles SAP_BR_BUYER).
c) Business Data:
A few required scope items such as Purchase Contract BMD, Consumable Purchasing BNX, Procurement of direct materials J45 need to be run and the corresponding data to be available. The key step here is the identification of any required scope items to be implemented and hence the data created accordingly.
d) Model training:
Finally train or re-train the model and activation of the required model version is to be done.
e) Access the app:
Now logon to the Fiori LaunchPad as the “Buyer” and access the app – Quantity Contract Consumption and follow the steps as specified in the scope item 1QR help documentation to run the scenario and see the predicted consumption results of the contracts to be expired.
The above 5 steps are to be done for any of the embedded predictive scenarios/use cases that were developed and released out-of-the-box with SAP S/4HANA functionality.
Let us now quickly review the embedded predictive scenarios that are released – with the scope item names, user roles required, Fiori IDs and any other pre-requisite scope items needed. With that understanding you would be more confident on how to proceed with your current implementation of embedding predictive functionality into SAP S/4HANA business processes.
Scope items in the embedded predictive scenarios
| Use Case | LoB | Scope item | User Role | Component | Fiori IDs | Status |
| SAP Tax Compliance Smart Automation / GRC | Finance | |||||
| Business Integrity Screening / GRC | Finance | SAP_BR_CASH_MANAGER | ||||
| Deprecated | ||||||
| Deprecated (new use case planned) | ||||||
| Contract Consumption | Procure | 1QR | SAP_BR_BUYER SAP_BR_PURCHASING_MANAGER SAP_BR_PURCHASER | MM-FIO-PUR-ANA | F2012, F1837 | |
| Deprecated | ||||||
| Supplier Delivery Prediction | Procure | 3FY | SAP_BR_BUYER SAP_BR_PURCHASING_MANAGER SAP_BR_PURCHASER | MM-FIO-PUR-ANA | F1837,F2358 | |
| Stock in Transit | Produce | 20N | SAP_BR_INVENTORY_MANAGER | MM-FIO-IM-SGM | F2139, F1837 | |
| Demand-Driven Replenishment: Dynamic Buffer Level Adjustment (using stock transfer) | Produce | 20N | SAP_BR_PRPDN_PLNR | PP-DD | F2831, F1837 | |
| Defect Code Proposal (incl. Text Recognition) | Produce | 20N | SAP_BR_QUALITY_TECHNICIAN SAP_BR_QUALITY_ENGINEER | F2649,F2868 | ||
| Early detection of slow and non Moving stocks | Produce | 20N | SAP_BR_INVENTORY_MANAGER | MM-FIO-IM-SGM | F2137 | |
| Quotation Conversion Probability Rate | Sales | 2YJ | SAP_BR_SALES_MANAGER | SD-FIO-HBA | F1904, F1871, F1837 | |
| Sales Forecast | Sales | 2YJ | SAP_BR_SALES_MANAGER | SD-FIO-HBA | F3304 | |
| Delilvery Performance / Delivery in Time | Sales | 2YJ | SAP_BR_SALES_MANAGER | SD-FIO-HBA | F3408, F1837 | |
| Sales Performance Prediction (formerly Sales Forecast) | Sales | 2YJ | SAP_BR_SALES_MANAGER | SD-FIO-HBA | F3304, F1837 | |
| Process Implausible Meter Reading Results | Utilities | |||||
| Process Outsorted Billing Documents | Utilities | SAP_BR_BILLING_SPECIALIST_ISU | F2186 | |||
| Reactive Maintenance | 4HH | |||||
| Behavioral Insights | Public Sector | |||||
| Business Rule Mining | MDM |
Now that we understood the steps involved in implementing the embedded predictive scenarios with an example and also briefly highlighted all the embedded predictive scenarios released across the different LoBs, let us now dive into the world of side-by-side ML scenarios.
Side-by-side Machine Learning Scenarios:
Let us now look into the Side-by-side ML scenarios that consume Machine Learning services from the SAP Business Technology Platform and how they are utilized by the SAP S/4HANA business processes. The following steps highlight the flow:
- System access – The system is accessible via the Fiori LaunchPad. The system administrator provides the URL to access accordingly, the various apps assigned to your role.
- Roles – The specific role for the ML service need to be assigned and should be used.
- Preliminary steps – Creation of master data, organizational data and other data needed for the ML scenario.
- Business Conditions – Any pre-requisite scope items need to be implemented first for the basic business conditions to be met.
- Configuration – Configure the ML service.
- Subscription – Subscribe to the corresponding application that uses the service, the scope item has the complete details of the service.
- Communication – Create the comm system as the SAP_BR_ADMINISTRATOR. Then create the COMM scenario assigned for the specific ML service.
- Training – Schedule the training job.
- Infer the results from the prediction models by changing the role and login as the specific end user to access the app and run the functionality to see the predictions.
Let us now take the example of a scope item 3NF – Machine Learning for Accruals Management. This also requires a pre-requisite scope item 2VB – Purchase Order Accruals. Here the accruals management provides recommendations during the accrual review process.
a) Scope item and flow:
3NF – The machine learning service used for accruals management is a Cloud service that uses machine learning technology to observe your accruals management and provide recommendations during the accrual review process. To support the process, the machine learning service can learn from decisions taken in the past, and apply learned knowledge to the new business situation. For accrual amounts that need manual review, the system adopts the machine learning service and then provides recommendations for reliable accruals for each purchase order. You can also review all the reliable accruals or only the reliable accruals that are above a certain confidence level in one go by using the mass review function.
b) Roles: You will need to start with the SAP_BR_ADMINISTRATOR role to do the required configuration.
c) Business Data and Pre-requisites for configuring the ML Service:
- The scope items 2VB (Purchase Order Accruals) and XX_3NF (Machine Learning for Accruals Management (Cloud only)) are both active.
You can check this in the appManage Your SolutionunderView Solution Scope.
If the scope item is not active, please request the activation via a BCP ticket on component:XX-S4C-OPR-SRV.
- The Accruals Recommendation service is active in your account on SAP Business Technology Platform (SAP BTP).
You can request the activation via a BCP ticket on component:CA-ML-OPS.
After the service activation you should be able to see the Accruals Recommendation service in the Cloud Foundry service marketplace, under any space in your BTP account.
To create a space, you can go to the activated subaccount, selectSpacesand clickNew Space.
d) Subscribe to the Accruals Application:
- Open the space in SAP Business Technology Platform.
- UnderServices, openService Marketplace.
- Choose the serviceAccruals Recommendationtile.
- To create a new service instance, chooseNew Instance.
- UnderService Keys, chooseCreate Service Key. The system generates and displays the oAuth credentials.
e) Create the communication system:
- Log on to the SAP Fiori launchpad as anAdministrator.
- Select theCommunication Systemstile.
- On theCommunication Systemsscreen, chooseNew.
- Make the following entries:
| Field | User Action or Values | Example |
| System ID | system ID | ACCRUALS_ML_INTEGRATION |
| System Name | system name | ACCRUALS ML COMMUNICATION SCENARIO |
- ChooseCreate.
- UnderTechnical Data, fill in the following fields:
| Name | Description |
| Host Name | The host name for target system. |
| OAuth 2.0 Endpoint | The endpoint of oAuth authentication server. |
| OAuth 2.0 Token Endpoint | The token endpoint of oAuth authentication server. |
- UnderUser for Outbound Communication, create a new user with the following information:
| Name | Description |
| Authentication Method | OAuth 2.0 |
| OAuth 2.0 Client ID | The client ID of oAuth authentication server user. |
| Client Secret | The client password of oAuth authentication server user. |
- ChooseCreate.
- ChooseSave.
f) Create the communication Arrangement:
- Log on to the SAP Fiori launchpad as anAdministrator.
- UnderCommunication Management, select theCommunication Arrangementstile.
- On theCommunication Arrangementsscreen, chooseNew.
- In theNew Communication Arrangementdialog box, in theScenariofield, enterSAP_COM_0446.
- ChooseCreate.
- TheCommunication Arrangementsdisplays.
- In theCommon Datasection, in theCommunication Systemfield, select the communication system that you created in the previous step:Create Communication System.
- ChooseSave.
g) Schedule the training job:
- Log on to the Web UI for your SAP S/4HANA system using the user you received.
- In theAccruals Managementbusiness group, openSchedule Accruals Job.
- ChooseNew.
- As a job template, chooseTrain Accruals Prediction Model on Historical Data.
- UnderScheduling Options, set the running schedule according to your requirement. The default frequency is set to one week.
- ChooseBackand monitor the background job.
h) Train the Accruals Prediction Model based on historical data:
- This functionality is available in theSchedule Accruals Jobsapp. Select theTrain Accruals Prediction Model on Historical Datatemplate.
- A machine learning service which is a feature of theReview Purchase Order Accruals – For Cost Accountantapp predicts whether user will adjust the proposed periodic planned costs. This job takes data from the table that contains the history of the previous interactions of the cost accountants and trains the prediction model using these data.
i) Infer Accruals from the prediction model:
- This functionality is available in theSchedule Accruals Jobsapp. Select theInfer Accruals from Prediction Modeltemplate.
- A machine learning service, as a feature of theReview Purchase Order Accruals – For Cost Accountantapp, helps to predict whether you need to adjust the proposed periodic planned costs.
- You run this job best outside of business hours after theTrain Accruals Prediction Model on Historical Datajob is finished.
Let us now quickly review the side-by-side ML scenarios that are released – with the scope item names, user roles required, Comm Scenarios and any other pre-requisite scope items needed. With that understanding you would be more confident on how to proceed with your current implementation of ML functionality for the SAP S/4HANA business processes.
Scope items in the side-by-side ML scenarios
| Use Case | LoB | Scope item | User Role | Comm Scenario | Status |
| Cash Application | Finance | 1MV | SAP_BR_CASH_MANAGER | SAP_COM_1018 | |
| Remittance Advice | Finance | 1MV | SAP_BR_CASH_MANAGER | SAP_COM_1018 | |
| Cash Application (Feature Release) | Finance | 1MV | SAP_BR_CASH_MANAGER | SAP_COM_1018 | |
| SAP Cash Application (Feature Release II) | Finance | 1MV | SAP_BR_CASH_MANAGER | SAP_COM_1018 | |
| Payment Advice Extraction (old name: Remittance Advices) | Finance | 1MV | SAP_BR_CASH_MANAGER | SAP_COM_1018 | |
| Goods Receipt / Invoice Receipt Monitor ML Status Proposal | Finance | 2ZS | SAP_BR_ADMINISTRATOR | SAP_COM_0246 | |
| Deprecated | |||||
| Intelligent Accrual Recommendation | Finance | 3NF, 2VB | SAP_BR_ADMINISTRATOR | SAP_COM_0446 | |
| Integrated Digital Content Processing for Content Mgt. | Idea | 2YC | SAP_BR_ADMINISTRATOR | SAP_COM_0245 | |
| Deprecated (new use case planned) | |||||
| Deprecated (new use case planned) | |||||
| Deprecated (new use case planned) | |||||
| Deprecated (new use case planned) | |||||
| Deprecated (new use case planned) | |||||
| Deprecated (new use case planned) | |||||
| Create Sales Orders from Unstructured Data | Sales | 4X9 | SAP_BR_INTERNAL_SALES_REP | SAP_COM_1129 | |
| Intelligent Intercompany Reconciliation | Finance | 4LG | SAP_BR_RECON_ACCOUNTANT | SAP_COM_0553 |
To help the customers and the partners, we also released a “Best Practices of doing Predictive Analytics and Machine Learning with SAP S/4HANA” in Q2 2021. You can find more information on the best practices at On-Premise (embedded and side-by-side), Cloud (embedded and side-by-side), explorative analytics.This would help to provide the documentation, technical guide set-ups and direct reference/access to all the released use cases – embedded ML, side-by-side ML and explorative predictive analytics.
We are also releasing a comprehensive book – “Implementing Machine Learning with SAP S/4HANA” by SAP-Press in the mid of September 2020.
Here are some quick links to the blogs in this series to give you a complete understanding of how Predictive Intelligence is infused into SAP S/4HANA.
- Resources and journey to machine learning with SAP S/4HANA
- Part 1 – Leveraging Predictive Intelligence with SAP S/4HANA
- Part 2 – Architecture and deep-dive of the different approaches around Predictive Intelligence
- Part 3 – Process flow leveraging Machine Learning and Predictive Analytics
- Part 4 – Scope and functionality in the context of an end-to-end process leveraging ML
- Part 5 – Activating machine learning functionality for SAP S/4HANA (this blog)
- Part 6 – Building ML into the digital core of SAP S/4HANA (Embedded ML)
- Part 7 – Enhancing the digital core with ML Services (Side-by-Side ML)
- Part 8 – Extending the digital core by leveraging ML with SAP Analytics Cloud
- Part 9 – ML Extensions to SAP S/4HANA processes
- Blog series – ISLM for machine learning with S/4
- Introducing the book – Implementing Machine Learning with SAP S/4HANA
Happy predicting the future!!
