Ein ansatz zur automatischen Umwandlung von Rohdaten in Regeln: Teil 1: Prozeßtrends, wavelettransformation und klassifizierungsbäume

The increasing amount of data which is measured in processes of many different branches is often not sufficiently exploited. Moreover, the increasing abilities concerning data collection are often not met by appropriate capabilities of making use thereof. Consequently, more emphasis should be put on the development of evaluation techniques, which allow to convert raw data into more meaningful information. In this article, a concept is introduced which automatically extracts regularities of a process series that indicate an unsuccessful outcome. These regularities are expressed in rules such as those used in expert systems.