Analysis of Order Data Customer Segmentation in Logistics Companies Using K-Medoids and DBSCAN Algorithms

Mujiono Sadikin (Program Studi Informatika Fakultas Ilmu Komputer Universitas Bhayangkara Jakarta Raya)
Nanda Azvita (Program Studi Informatika Fakultas Ilmu Komputer Universitas Bhayangkara Jakarta Raya)

Technologia Journal
Vol. 2 No. 4 (2025): Technologia Journal-November

The development of the logistics industry makes the use of customer data to understand market behavior and needs increasingly important. This study aims to segment customers based on logistics company order data using the K-Medoids algorithm and Density-Based Spatial Clustering of Applications with Noise (DBSCAN). This approach is used to identify customer groups with similar characteristics to support more effective marketing and service strategies. This study uses 12,000 customer order data entries from the past year, with variables including order, cost, and receiving location. The data is processed through preprocessing stages (cleaning, transformation, and normalization) before being applied to two clustering models. The analysis results show that the K-Medoids algorithm produces a Silhouette Score of 0.3559, while DBSCAN obtained a score of 0.3233. These values ​​indicate that K-Medoids has more compact and well-separated clusters than DBSCAN. Thus, the K-Medoids method is more effective in segmenting customers to support strategic decisions of logistics companies.