A Spatially Constrained PCA–MST-Based Clustering Model for Railway Freight Management in Kazakhstan

Background: Kazakhstan’s railway network exhibits substantial spatial heterogeneity, limiting the effectiveness of uniform freight management strategies and necessitating differentiated analytical approaches to regional transport planning. This study aims to develop a spatially constrained clustering model for railway freight management based on the economic, infrastructural, and operational characteristics of the regions served by KTZh—Freight Transportation LLP. Methods: The proposed methodology integrates principal component analysis (PCA) with a minimum spanning tree (MST) algorithm under railway connectivity constraints. A dataset comprising 20 standardized indicators for 17 regions of Kazakhstan was analyzed. Results: PCA reduced the original variable space to five principal components, explaining 77.9% of the cumulative variance. Cluster validity was confirmed using the Elbow, Silhouette, and Calinski–Harabasz indices, resulting in the identification of six spatially connected transport clusters with distinct functional profiles. The clusters revealed significant regional differences in freight generation, logistics infrastructure, transit potential, and investment characteristics. Conclusions: The proposed framework provides an evidence-based analytical tool for railway freight management, infrastructure planning, and the prioritization of regional development strategies while accounting for spatial connectivity constraints.