Mathematics revolutionizes the supply chain

Digitalization is changing the face of supply chain management. Technological innovations, in combination with the latest methods of data analysis, now offer more and more opportunities to optimize processes and develop new services.

Alongside advances in these methods, recent years have also brought major improvements in the performance of numerical solution algorithms, thereby making it possible to rapidly process a substantially greater volume of data with direct relevance to the application.

In the past, technological and methodological constraints meant there was a focus on descriptive analytics, whereby historical values are analyzed and interpreted. Today, descriptive analytics has been joined by predictive and prescriptive analytics. Their purpose is to predict the future as precisely as possible and to optimize processes. The use of methods, procedures and models from mathematics and statistics now makes it possible to automatically analyze procedures and events – and, on this basis, to derive concrete measures for process planning and control.

One of the key problems in supply chain management is that logistics processes continuously generate massive volumes of data – shipment data, data from warehouse management systems, data from machine sensors, etc. In most cases, this data is recorded and saved for a specific task – for management, information or documentation purposes.

In practice, problems frequently arise when using modern methods of data analytics to analyze and evaluate this data. This is because the data is often erroneous or patchy or otherwise not of the required quality. Furthermore, in some cases, the data needed for a specific application is unavailable, as it is either impracticable or too costly to collect it. In other words, it is often the case that the available data does not meet the specifications required for the algorithms.

Optimal value-creation processes can be achieved only with data analytics  

Therefore, in such an environment, how can we best harness data analytics in order to develop predictive systems that can identify problems in advance, recommend a suitable course of action and, wherever possible, automatically trigger an ideal response?

The answer is to combine forecasting and optimization methods and thereby reduce the degree of complexity in mathematical representation. Fraunhofer SCS is therefore expanding existing data structures and combining them with existing data analysis methods in order to develop new solutions. We select, combine and adapt procedures and algorithms in line with the requirements of the specific application. This delivers solutions that can monitor logistics chains, forecast key indicators and events in the supply chain, and plan and control the supply chain right up to and including personnel deployment.