Optimizing MSMEs clusters using type-2 Fuzzy K-Medoids method in Sampang Madura District

MSMEs in Madura, Indonesia, play a crucial role in driving local economic growth, yet face significant challenges such as limited capital, disparity in market information, and low technology adoption. Data disparities among MSMEs, such as disproportionate differences in assets and production capacity, indicate uncertainty in data characteristics. The research proposes the use of Fuzzy K-Medoids Type-2, which is capable of handling data ambiguity and uncertainty through fuzzy membership representation. The study using CRISP-DM approach, starting from business understanding. The data used included 1,275 MSMEs with five main variables: number of employees, production capacity, turnover, assets, and business licenses. The evaluation was conducted by comparing the Silhouette Coefficient (SC) for K-Medoids and Partition Coefficient (PC) for Fuzzy K-Medoids Type-2. The results show that the optimal configuration is obtained with a value of k = 2 in Fuzzy K-Medoids Type-2, with a PC of 0.9548. The first cluster is categorized as small MSMEs with a low average variable, while the second cluster is categorized as medium MSMEs with a higher average variable. These findings provide a strong basis for developing a more targeted and data-driven MSME development strategy.