Kombiniertes Data Mining - Klassifikation unter Verwendung von durch Clustering gewonnenen Hilfsinformationen

Problems in data mining are versatile. In some cases they can only be solved by combining different data mining methods. Existing approaches concerning a combined use of data mining methods consider that topic under an isolated point of view. This work continues a parallel study [SK04] and further introduces the term “Combined Data Mining”. Through combination of data mining methods the result of the combined process should gain quality and efficiency by letting the used methods interact with each other. One possible way to achieve interaction is to compute additional information in a preliminary step which is used in a succeeding step. In the course of this work clustering and classification become combined. Therefore a “decision tree classifier” is implemented, which uses by a clustering algorithm previously identified and computed additional information. Priority objective is to investigate additional information that can simply be applied to the classifier and has impact on the quality of the classifier.