Clustering Madura tourism destinations using Fuzzy C-Means and Fuzzy C-Medoids with Xie-Beni optimization

Abstract

Tourism is a strategic sector that plays a significant role in improving the regional economy, including Madura Island, which boasts diverse cultural and natural tourism destinations. Proper clustering of tourist attractions is essential to support more effective regional development planning, promotion, and policy strategies. This study applies data mining with an unsupervised learning approach to cluster tourist attractions in Madura using a comparison of two methods: Fuzzy C-Means (FCM) and Fuzzy C-Medoids (FCMedoids). Both methods were evaluated using the Xie-Beni (XB) validation index as an optimization parameter to determine the quality of cluster formation. Preprocessing included data normalization and outlier removal to ensure model stability. Experiments were conducted with varying the number of clusters from 2 to 10 to obtain the smallest Xie-Beni Index value as the best clustering result. The results showed that the Fuzzy C-Medoids method produced a lower Xie-Beni value of 0.09 compared to the Fuzzy C-Means method (5.40), indicating better separation between clusters and a higher density within clusters. These clustering results can be used as a basis for decision-making in developing regional tourism potential, data-driven promotional strategies, and planning sustainable tourism policies on Madura Island.