Management Science & Business Administration

Prescriptive analytics techniques harness data to make decisions, such as how much inventory to stock, which orders to produce in the coming week, or how to route delivery vehicles. We specialize in modeling large-scale, real-world optimization problems and solving them using modern optimization methods from the field of operations research, such as mixed-integer programming and metaheuristics, to find optimal decisions. In addition, we make use of machine learning methods, such as deep neural networks and decision trees, to assist optimization methods to find good solutions quickly.

We have modeled and solved a wide variety of problems in the area of prescriptive analytics, such as for designing shipping networks, scheduling and routing technicians, planning the maintenance of vehicles, inventory management for grocery retailing, and employee shift scheduling, among many more. We further specialize in finding solutions that are robust to variations in input data, such as due to variations in travel times or demand. Overall, our solutions help businesses lower energy consumption and more effectively use their available resources.

Within the broad range of interesting topics in the field of innovation and technology management, we currently scrutinize research questions regarding (*) economic implications of smart products and services (e.g., corresponding business model innovations, smartness barrriers), (*) market diffusion of innovations and technologies (e.g., planning the market introduction of new products, multi-generation technology succession), (*) sustainability issues (e.g., promoting respective innovations), (*) cooperation between market players in developing new products or services (e.g., between an incumbent firm and a startup company or between universities and industry), (*) management of research and development (e.g., R&D project selection and resource allocation), and (*) digital platforms (e.g., measures for overcoming the chicken-and-egg problem when establishing a two-sided platform).

In doing so, we apply various methods (or combinations thereof). Most prominently, we use agent-based market simulation either as a means to study the above research questions or to further theory building (e.g., with respect to the effects of stakeholders’ heterogeneity and individuality regarding preferences and behavior). Furthermore, we use quantitative empirical studies (most often, following an experimental design), qualitative empirical studies (most often, interview studies or case studies), and mathematical modeling of real-world problems in order to provide (interactive) decision support.

Our research activities and competences primarily lie in the area of quantitative consumer behavior analysis, with a special focus on social media, and in the area of model-based decision support for smart products and services. We therefore use, and partly further-develop, econometric/stochastic methods as well as methods and algorithms from machine learning. The data basis for our research stems, for example, from (adaptive) online experiments on the one hand and from extensive internet crawling on the other. The contents of these databases range from brand-related word-of-mouth to new product-related individual preferences. In the field of analyzing large amounts of marketing data, we can also draw on extensive experience in problem-centered (further) development of neural networks (e.g., multilayer perceptron and neural gas). Our research is also closely related to the field of customer experience management and model-based brand image analysis. As researchers interested in the practical relevance of their empirical findings, it has always been a concern to successfully apply the methods and principles that have proven to be effective in research settings to marketing practice.