Customer Relationship Management technology groups customers based on their transaction details. The focus of the paper is finding groups of clients with similar buying patterns in a department store using these data for targeted marketing. The performance of K-means, K-Medoids, Agglomerative Clustering and DBSCAN is compared on a real retail dataset of a department store. Results show that the K-Medoids algorithm is more robust in case of noise and outliers in data. K-Means and DBSCAN perform better in terms of time, especially for large retails datasets.
Ledion Lico
Indrit Enesi
Betim Cico
Aba Business Centre
Department of Electronics and Telecommunications – Polytechnic University of Tirana
Department of Computer Science – Metropolitan University of Tirana
Albania
e-mail: ledion.lico@fti.edu.al
Abstract:
Key words:
Agglomerative
DBSCAN
cluster techniques
K-means
K-Medoids
Section: