Research on Landscape Architecture Space Optimization Based on Machine Learning: Taking Tourist Behavior Analysis as an Example

This study presents a data-driven framework integrating machine learning (ML) techniques to analyze tourist behavior and optimize spatial configurations in landscape architecture. By leveraging GPS trajectory data, IoT sensor networks, and GIS spatial analytics, we develop models to cluster visitor groups, predict activity preferences, and forecast crowd dynamics. Key methodologies include DBSCAN for activity zone identification, Random Forest for spatial preference modeling, and LSTM networks for congestion forecasting. A case study of Qingfeng Park in Shanghai demonstrates the framework's application, yielding a 32% reduction in peak-hour congestion and a 28% improvement in facility utilization. Mathematical formulations for spatial interaction modeling and model evaluation metrics are explicitly detailed, providing a replicable framework for evidence-based landscape design.