Data in, insights out
Pandas and Python are familiar from the animal kingdom, but also from the data jungle. There, they aren’t exotic creatures, but rather common tools. Anyone working with modern analytical methods is just as familiar with them—as a suitable programming language or program library—as they are with visualization tools that originally came from the video game industry. Take Unreal Engine, for example, which once transported the gaming community to the distant galaxies of the Star Trek universe. Or Panda 3D, which brings Pirates of the Caribbean to the screen after initially bringing comic book characters to life in Disney theme parks.
So are the experts at the general planning firm io also at home in the tropics, in space, on the high seas, and in Duckburg? “When it counts: yes,” admits Dr. Jens Koenig, Principal Consultant at io, with a laugh. Because whether it’s AI and statistics, logistics and supply chain analytics, optimization and simulation, or operational dashboards: for a company to take the right growth measures, operational data must be extracted and processed, results interpreted, and decisions made.
“Nevertheless, many organizations still rely on opaque Excel spreadsheets that don’t allow for comparability, require manual intervention, and turn data processing into a tedious, time-consuming task,” Koenig reveals. Given the wide range of powerful analytics tools available, this approach is no longer appropriate. This becomes all the more evident with the integration of analytics with AI and SAP Connectors—a standard practice among forward-thinking planning service providers.
An SAP Connector is based on the OData service. As an interface that connects external applications to SAP systems, it can combine data from the ERP system and SAP EWM and integrate additional modules. This makes it possible to build analytical models from which insights can be derived and visualizations created. In addition, data can be fed back to SAP via the SAP Connector, although—depending on the setup—other feedback channels such as text files, XML, or JSON can also be used.
io, for example, uses the SAP Connector to create analytical models for its customers—for which, ideally, millions of individual data points are provided. The SAP Connector can automatically retrieve the required data from the SAP system—which significantly reduces the workload. It would be impossible for individual employees to process millions of data records. On top of that, the necessary access permissions are often lacking—for example, when a company has multiple locations. The SAP Connector acts, so to speak, as an interface through which data can be exported—and the recipient does not need an SAP system itself. It is also possible to feed data back via the SAP Connector, for example, to update master data for bulk data.
AI, for its part, has long since established itself as a reliable tool for forecasting, optimization, and simulation models, as well as for pattern recognition, classification, and clustering. That said, it’s hardly accurate to speak of “one” AI. There are language models such as GPT-3, which focus on understanding and generating natural language. They use deep neural networks to learn the structure and meaning of language. Analytics AI, on the other hand, focuses on the analysis and interpretation of existing data. It uses algorithms and machine learning to identify patterns, make predictions, and develop optimizations. In addition, there are hybrid models that combine the advantages of both approaches.
For the deep dives that Koenig and his colleagues conduct as part of client projects, they always rely on the most cutting-edge tools and methods. The practical examples presented below illustrate what this entails. Finally, the topic is illuminated by a look into io’s in-house research and development.
