Featured Projects

With the speed in which IT topics are now being pushed forward, even media companies have to adapt more quickly and become flexible. In particular, this includes marketing and sales activities in order to keep the customers satisfied and to raise further potential. For the analysis project of the Smart Data Solution Center Baden-Württemberg (SDSC-BW), the Huber publishing company provided anonymised information about the concluded contracts of their services. The contracts and customer data, as well as the related marketing activities, were collected over a period of 72 months. In total, information from 943 database tables was processed and a quarter of a million data sets were analyzed and evaluated.

Over several years, the KIT-FM (Facility Management) has collected data with immense value for the operational management, but also for the planning and implementation of future infrastructure developments. This data is also of great interest for researchers. On the one hand, we will examine how the existing infrastructure data evaluated by Smart Data methods can help to draw more accurate conclusions about the operational management and the infrastructure planning. On the other hand, we will drive forward the usability of this data for research and innovation projects.

The project mainly aims at the reduction of the rework and the avoidance of errors during the production of tractors at the John Deere factory in Mannheim. These two objectives are realized through a data analysis of the error information, the test protocols and their interdependencies. Based on the results of the data analysis, we can make prognoses and rules for the production planning that help the company to take one step further in the process of self-optimization.

Predicting shipping volumes with artificial intelligence instead of intuitive prediction was the goal of the smart data experts at SDSC-BW together with the logistics and transport company LGI. Various algorithms were implemented, for daily, weekly and monthly prediction, and evaluated in order to find the best model. The complex models of SDSC-BW could significantly outperform the prediction models in the comparative analysis.

The churn prediction is an important method to predict customer churn through machine learning and data mining. The challenge is to enable companies with a precise and real-time prediction, giving them enough time to keep their customers. Previous research in the B2B-context, as well as in the B2C-context is missing the dynamic aspect of macroeconomic variables in the time elapsed. It is the goal of this work to create a churn prediction model with the use of machine learning algorithms like Random Forest and neural networks and to research if an inclusion of dynamic aspects will lead to improvement.