Disaster resilience often overlooks the suitability of schools and community shelters, leading to uneven safety outcomes during emergencies. This research addresses that gap by developing a data-driven shelter suitability prediction model using Gaussian Process Regression (GPR). The model integrates key urban parameters such as environmental risk factors, infrastructure stability, and population density to predict shelter safety scores across the city. These scores are then visualized spatially to identify safer zones, schools with better access to open spaces, emergency resources, and lower hazard exposure. Conversely, low-scoring areas highlight regions at elevated risk, guiding authorities toward targeted reinforcement and resource allocation. Outlier detection techniques further refine the analysis, pinpointing schools with unusually high or low suitability for deeper investigation. The model’s performance, evaluated through five-fold cross-validation, reveals variability in Mean Squared Error (MSE) across folds, suggesting the potential for ensemble-based optimization. By coupling predictive modeling with geospatial visualization, this study provides a powerful decision-support framework for urban planners and disaster management authorities to prioritize structural improvements and evacuation planning, enhancing community resilience before a crisis strikes.