Fuzzy-Clustering als methodische Grundlage zur dynamischen Marktsegmentierung
Anneke Minke · HilDok – Institutional Repository (Universität Hildesheim) · 2026
Not cached yet. This record was fetched live for this request and is stored after the lookup. Reopening the page should serve the cached copy.
HilDok – Institutional Repository (Universität Hildesheim) · Springer Nature (Switzerland) · dissertation · 2026
Abstract
In market research problems, detecting time-dependent changes in data structures is of essential importance for systematic future planning.Especially understanding changes concerning separate market segments is a fundamental task for cooperations in order to react to specific developments and to initiate appropriate actions.However, an exclusively descriptive analysis of existing structural changes is not sufficient, but instead emerging trends have to be included in planning strategies to modify different marketing tools for promoting certain developments or -if possible and reasonable -to counteract them.The field of Change Mining is concerned with approaches for analyzing structures in a dynamic environment and hence, to detect time-dependent changes at an early stage.Cluster analysis is a widely used technique for identifying market segments automatically as it enables dividing a heterogeneous data set, e.g.data regarding customer behavior, into homogeneous subsets.Additionally, when analyzing time-dependent changes within a data set, the detection of gradual differences between different points in time is essential for predicting future developments and trends.Therefore, hard clustering techniques are of limited use as membership degrees are needed rather than crisp cluster assignments.Hence, fuzzy clustering appears suitable as basis for the dynamic examination of market segments.In this work, a general introduction to Data Mining and Change Mining is given and on this basis, Data Mining approaches for analyzing time-dependent changes are presented which are relevant for different market research tasks.Afterwards, the fundamentals of fuzzy clustering are introduced.Due to the focus on detecting changes in a general cluster structure, new approaches for possibilistic fuzzy clustering are defined which emphasize the homogeneity of individual clusters and which are independent of possible changes that may occur over time.For tracking emerging changes and predicting potential developments, different measures are introduced and evaluated for demonstrating the Change Mining process.Thereby, gradual changes within individual clusters and abrupt changes concerning the overall cluster structure are considered separately.The results concerning gradual changes with regard to size, density, and volume of particular clusters are consulted when investigating abrupt changes relative to the number of clusters, i.e. newly developed, eliminated, merged, or split clusters, as they can indicate abrupt changes.The analysis of these types of changes is demonstrated using detailed illustrations of experimental results.The experiments were conducted by employing a Java applet which was exclusively developed for this purpose.
Open the resolved locator · 10.25528/368
Metadata licence: OpenAlex metadata (CC0)