<![CDATA[Road damage on access routes to coastal tourism areas in Tanah Laut Regency, South Borneo has become a critical issue affecting travel safety and tourist visits. Various types of pavement distress such as cracking, depression, bump and sags, patching and potholes, polished aggregat, rutting, and swelling create complex data patterns that require robust analytical methods. This study adopts a data-driven approach to compare the performance of three clustering algorithms K-Means, Hybrid PSO–KMeans, and the Butterfly Optimization Algorithm (BOA) to determine the optimal grouping structure of road damage data. The dataset consists of seven types of road distress obtained from field surveys across three coastal locations. Data preprocessing was carried out through normalization and standardization to ensure consistency in scale, followed by clustering analysis with varying numbers of clusters (k = 2 to 7). The Silhouette Coefficient was used to evaluate clustering performance and determine the optimal number of clusters. The results show that the optimal clustering structure is achieved at k = 3, representing three levels of road damage severity: minor, moderate, and severe. Among the evaluated methods BOA produced the highest Silhouette Score of 0.7559, outperforming Hybrid PSO–KMeans (0.6583) and K-Means (0.442) indicating more compact and well-separated clusters. These findings suggest that BOA is more effective in handling complex and heterogeneous road damage data. Practically, this approach can support data-driven decision-making in prioritizing road maintenance.]]>
