<![CDATA[Polling-station supervision in local elections requires alignment between supervisory capacity and field-level demands; however, polling station supervisors are often treated as a homogeneous workforce despite significant variations in characteristics and working contexts. This uniform approach may lead to policy–capacity mismatches, including inappropriate task allocation, uneven supervision quality, and potential bias in performance evaluation, which can affect electoral integrity. Despite growing attention to electoral governance, data-driven approaches in managing supervisory human resources remain limited, with practices still dominated by administrative and uniform models. This study addresses this gap by applying a data-driven segmentation framework using recruitment data from 2,197 polling station supervisors in the Bangka Belitung Islands, integrating Market Basket Analysis (MBA), Principal Component Analysis (PCA), and K-Means clustering. The results identify three distinct profiles: (1) younger supervisors with limited experience requiring technical support, (2) a highly diverse segment requiring flexible and context-specific management, and (3) more stable mid-aged supervisors with stronger adaptability. These findings demonstrate that polling station supervisors cannot be managed uniformly; instead, segmentation enables differentiated policy design, including targeted training, more precise deployment, and fairer performance evaluation. This study contributes by advancing data-driven governance in electoral supervision and bridging data analytics with public policy in election management.]]>
