OPTIMALISASI DISTRIBUSI PEMILIH TERHADAP TPS MENGGUNAKAN METODE CLUSTERING FUZZY C-MEANS

Abstract

General elections are a fundamental pillar of modern democratic systems, requiring an implementation that is efficient, fair, and inclusive. One of the key factors influencing the success of an election is the determination of polling station locations, as their placement directly affects voter accessibility, travel distance, and public participation. Inappropriate polling station allocation can lead to service inequality, voter congestion, and a decline in the overall quality of the voting process. At the local administrative level, polling station determination is still largely conducted manually by grouping voters based on neighborhood or administrative boundaries. This conventional approach is often time consuming, prone to administrative errors, and frequently results in an uneven distribution of voters across polling stations. In addition, electoral regulations impose limits on the maximum number of voters per polling station to ensure smooth and orderly voting procedures, which are not always optimally satisfied through manual methods. As voter data complexity and geographic dispersion increase, computational approaches are needed to support more effective decision making. Clustering techniques in unsupervised learning enable objective grouping of voters based on spatial characteristics. The Fuzzy C-Means method represents a suitable approach because it can accommodate data uncertainty and overlapping service areas. The application of this method is expected to produce a more efficient, equitable, and data driven distribution of polling stations, thereby contributing to the improvement of election management quality and democratic integrity