Design and Implementation of a Big Data Decision Support System for Balanced Allocation of Preschool Education Resources

To address the dynamic imbalance between preschool education demand and resource allocation under rapid urbanization, this study proposes a spatiotemporal big data decision support system integrating multi-source data fusion, demand prediction, spatial accessibility assessment, and multi-objective optimization. The system adopts a front-end/back-end separation architecture and combines a Spatio-Temporal Graph Convolutional Network (STGCN), an improved Two-Step Floating Catchment Area (i2SFCA) method, and Multi-Objective Particle Swarm Optimization (MOPSO) to establish an automated workflow for demand forecasting, mismatch diagnosis, and resource configuration. WebGL- and WebGIS-based visualization modules enable interactive analysis and real-time decision support for large-scale spatial data. Experimental results on real urban datasets demonstrate that the proposed framework reduces the demand forecasting mean absolute percentage error to 6.8%, decreases the global spatial accessibility Gini coefficient to 0.28, and improves the supply-demand matching rate by 23% under the optimized allocation strategy. The proposed architecture provides an efficient engineering solution for heterogeneous information processing and large-scale spatial optimization, offering methodological references for intelligent sensing networks, distributed information integration, and data-driven decision support in electromagnetic and communication-related infrastructure systems.