Step 4: Training

Objective

After completing this lesson, you will be able to execute the Training Process of the SAP Cash Application

Training Process of the SAP Cash Application

Overview on Jobs

The following steps have to be executed in order to train a model with SAP Cash Application for Contract Accounting:

Job	Transaction
Cash Application: Transfer Payment Lot Items and Trigger Training	FPML_CASHAPP_TRAIN
Cash Application: Query Training Status	Transaction FPML_CASHAPP_TSTAT

Cash Application: Transfer Payment Lot Items and Trigger Training

The selection screen and result screen with processing the training job in Cash Application. Also a subsequent process in the cloud is triggered when the back end process has been finalized on the customers S/4HANA system.

Cash Application: Transfer Payment Lot Items and Trigger Training - Transaction FPML_CASHAPP_TRAIN

A scheduled job retrieves historical payment lot items and pushes them into a training dataset provider API to enable the training.

When the data volume of the training data is high, several CSV files are sent to the Machine Learning server for each run date and run ID. In this case, you get back the relevant status for each CSV file, represented by the job ID.

FREQUENCY: There is no need to update the training model onan a certain frequency. You will run FPML_CASHAPP_TRAIN again for example, if additional company code will be live or there are major changes in business which may impact SAP Cash Application. If the later processed Inference Run still provides a high level of matches, there is no need to re-train the data model.

Cash Application: Query Training Status

The Training Status is shown. With this job you will check the status of the subsequent job in the cloud for successful processing.

Cash Application: Query Training Status - Transaction FPML_CASHAPP_TSTAT

Like we have for the Master Data run and its related check for status with FPML_CASHAPP_MDSTAT, for the training job we also have the need to check for the subsequent processing in the BTP system. This is done with the FPML_CASHAPP_TSTAT. A scheduled job can query the status of training for the Machine Learning server. You query the status for the date and run ID of the respective Transfer Payment Lot Items and Trigger Training job. If you will get status ‘80’, the training model has been created successfully.

The Different Jobs Running for the Training Process

The Different Jobs Running for the Cash Application Process

Different API’s processed in cash application. There is only a request from back end system and cloud system will provide a response. There is no call initiated from cloud to customers back end system.

This slide shows a summary of the different jobs running for the training process and what they do.

The selection categories get extracted out of the customizing and sent to the Machine Learning API. The same happens with master data like Business Partners, Contract Accounts and Provider Contracts. The difference here is that the master data status gets requested to make sure that the data is always up to date.

Also, the clarified items as well as the payment lot items get extracted and trigger the training process. Once the training has taken place, the training status can be fetched.

Next lesson