Machine Learning Cockpit – What next?
Hi, all absolvents of openSAP course on Machine Learning Cockpit for SAP Business ByDesign (MLC1, MLC2) and all enthusiasts of MLC in general.
Purpose of the blog is to discuss new ideas on how to utilize MLC to its fullest, or to share what other use cases could be setup and how. May you have any question or proposal, feel free to post them in Comments.
Let me start the blog by elaborating on question of Feature Enrichment.
The MLC Scenarios relies on Business Analytic data sources. While there more than 200 basic data sources provided out-of-the-box they are not necessarily suited for machine learning purposes. Either they have too little fields, or the fields are not particularly useful for finding their relationship to a Target Field (ML model training). Or there is no Target Field in a form usable by MLC. Example of the last case are Payment Delay and Delivery Delay use cases.
One possible solution to these types of problems is to export the data, process them in Excel, import them back as cloud data source and join original scenario data source with this new cloud data source.
Another possible solution is to utilize ODATA capabilities of SAP Business ByDesign and automate the whole the process. But with help of ODATA export/import interfaces together with aggregation capabilities of Excel we can also calculate completely new features, like for an example Customer Payment habits indicators. Having these new features included on Customer master data can make Customer data source much more ML relevant.
Please stay tuned for more details. Thank you.