Abstract:

Hierarchical and nonhierarchical clustering algorithms are applied in this paper in order to classify customers in one luxury goods company according to the relevant clients’ attributes in an SQL Server database. After conducting a hierarchical cluster analysis, dendrograms are used to visualize customer groups, which is also useful for initializing the nonhierarchical K-means clustering algorithm. The K-means algorithm is performed so as to determine the partition of the objects in several groups or clusters, so that the objects in one cluster are more similar to each other than they are to the objects in the other, different clusters. In this research study, the clusters are initialized by arbitrarily selecting one object per cluster to represent each cluster, whereas each of the remaining objects is assigned to one of the clusters according to the clustering criterion and calculating the distance between each object and each cluster center. The Euclidean distance, the cosine distance and the Minkowski distance are used during the calculation of the similarities. The separation distance between the resulting clusters is analyzed and illustrated using the Silhouette index. The model presented in this paper is validated for five clusters, whereas the analysis and the results might be used to develop market segments, identify repetitive behavior or trends in order to evaluate customer actions and develop marketing strategies, plans and loyalty programs.