Development of a Hybrid Fuzzy Geographically Weighted K-Prototype Clustering and Genetic Algorithm for Enhanced Spatial Analysis: Application to Rural Development Mapping

Introduction/Main Objectives: Clustering methods are crucial for geodemographic analysis (GDA) as they enable a more accurate and distinct characterization of a region. This process facilitates the creation of socio-economic policies and contributes to the overall advancement of the region. Background Problems: The fuzzy geographically weighted clustering (FGWC) method, which is a GDA technique, primarily handles numerical data and is prone to being stuck in local optima. Novelty: This study proposed two novel clustering methodologies: fuzzy geographically weighted k-prototypes (FKP-GW) and its hybrid clustering model, which combines genetic algorithm-based optimization (GA-FKP-GW). Research Methods: This research conduct simulation study comparing two of the proposed clustering method. For the empirical application, this study applied clustering technique using the official Village Potential Survey of Temanggung, Indonesia. Finding/Results: The evaluation results of experiments conducted on simulated data and study cases indicate that the proposed method yields distinct clustering results compared to the previous method while being comparably efficient. The empirical application identifies four distinct groups from the clustered villages, each displaying unique characteristics. The results of our research have the potential to benefit the development of the GDA method and assist the local government in formulating more effective development policies.