Answers are grounded in this paper and cite their sources.
Ensuring safety and serviceability in urban buildings has become increasingly challenging due to growing project complexity and operational uncertainties. This study proposes a data-driven, performance-based architectural framework for predicting the Safety Performance Index (SPI) and Serviceability Performance Index (SePI) using a hybrid PSO–XGBoost approach. A Building Performance Dataset was preprocessed through missing value treatment, categorical encoding, outlier removal using the IQR method, and feature normalization. Feature engineering captured structural intensity, construction efficiency, and quality-defect relationships, while Particle Swarm Optimization (PSO) selected the most informative features by minimizing RMSE and maximizing R². The optimized features were used to train an XGBoost regression model for SPI and SePI prediction. Experimental results demonstrated excellent predictive accuracy, achieving an R² of 0.9999 for both indices. Error metrics remained minimal, with SPI yielding RMSE = 0.009988 and MAE = 0.00784, and SePI achieving RMSE = 0.00842 and MAE = 0.00669. The proposed framework enables reliable performance assessment for urban building projects.
No takes yet. Share an insight, caveat, or question.
Miaomiao Zhang (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: