Extracting knowledge and computable models from data - needs, expectations, and experience

In modern industrial manufacturing, a great amount of data is gathered to monitor and analyze a given production process. Intelligent analysis of such data helps to reveal as much information about the production process as possible. This information is most useful if it is available in the form of interpretable and predictive models. Such models can be generated from data by means of (fuzzy logic based) machine learning methods. In this contribution we will describe industrial applications in the areas of process optimization and quality control where we have successfully established machine-learning methods as intelligent data analysis tools. Based on these applications, we will report the characteristics of machine-learning tools which according to our experience support successful applications in an industrial environment. Furthermore we describe some methodical aspects resulting from this applications.