<![CDATA[Ancol is the largest recreational destinations, attracting visitors from diverse backgrounds. However, in 2024 the company experienced an 11.96% decline in visitor numbers. This condition highlights the urgent need for more accurate customer segmentation to support targeted and effective marketing strategies. Accordingly, this study investigates whether a Multiplex Leiden can produce coherent visitor segments, while also examining the relative contribution of each layer to community formation. Unlike prior multilayer segmentation studies, this study leverages the Multiplex Leiden algorithm, which guarantees well-connected communities and has been shown to achieve higher modularity. This is among the first applications of Multiplex Leiden for visitor segmentation, offering improved community coherence and interpretability in a multi-layer behavioral network. To balance network structures and reduce cross-layer density bias, kNN backbone preprocessing was applied before community detection. The results reveal 18 distinct visitor communities with substantial variation in size. Layer-wise quality analysis shows that the socioeconomic status layer contributes the strongest influence on the detected communities, followed by spending behavior and experiential preferences. The clustering quality was evaluated using multiple metrics. An Adjusted Rand Index (ARI) of 0.617 indicates a stable, non-random visitor segmentation, while a positive total quality score of 1.086 reflects strong cross-layer community structure. A mean conductance value of 0.548 suggests moderately well-separated yet realistically overlapping communities. Overall, the findings empirically confirm the effectiveness and interpretability of the Multiplex Leiden algorithm with backbone preprocessing for visitor segmentation in multi-layer networks. Future research may extend this framework by incorporating additional behavioral or temporal data.]]>
Copyrights © 2026
