Fuzzy Reduced K-Means for analyzing high-dimensional social phenomena

As societies face rapid transformations, understanding complex social, economic, and demographic patterns has become increasingly important for policymakers and researchers. This paper introduces a novel statistical method designed to enhance the analysis of global socioeconomic phenomena by combining in a simultaneous methodology fuzzy clustering with dimensionality reduction. The proposed methodology is named Fuzzy Reduced K-Means and it addresses the limitations of traditional fuzzy clustering, which does not perform dimensionality reduction, by identifying latent dimensions that best explain the data structure in clusters. Additionally, the method offers high flexibility by allowing data units to belong to multiple clusters with varying degrees of membership, thus reflecting the complexity and overlap inherent in real-world socioeconomic phenomena. Through three real-world applications (one considering the benchmark Iris dataset and the other two related to social phenomena), this paper shows that the model is a valuable tool for research and policy development, since it offers a nuanced understanding of complex systems and enables informed decision-making, fostering a better grasp of the multidimensional nature of socioeconomic challenges.