Xgboost Modeling To Predicting Land Use-Based Trip Generation: Literature Review

Abstract

Reliable trip generation prediction is essential for transportation planning, traffic forecasting, and sustainable urban development. However, conventional statistical models may struggle to capture nonlinear relationships between land-use characteristics and travel demand. This study systematically reviews research published between 2019 and 2025 on XGBoost applications in land-use-based trip generation and related travel-demand prediction. A total of 25 studies were analyzed based on data characteristics, land-use and built-environment variables, modeling approaches, validation techniques, evaluation metrics, and research findings. The review indicates that XGBoost can capture complex nonlinear relationships and interactions among explanatory variables, making it a promising approach for modeling travel patterns. However, predictive performance varies across datasets, geographical contexts, target variables, and evaluation methods, limiting direct comparisons among studies. Key challenges include model interpretability, data quality and availability, spatial transferability, and integration with Geographic Information Systems (GIS) and real-time transportation data. This review synthesizes methodological developments, identifies limitations in the existing evidence, and highlights opportunities for improving model explainability, validation, and cross-context generalization. The findings provide a basis for developing more reliable and context-sensitive trip generation models to support evidence-based transportation planning and smart city applications.