Spatiotemporal trajectory analysis of human activities based on a parallelized Random Forest algorithm

The exponential growth of large-scale spatiotemporal data has created new opportunities for understanding human activity patterns and addressing urban challenges. However, the high dimensionality, nonlinearity, and massive volume of such data present significant computational challenges for traditional analytical methods. This study proposes a scalable parallelized Random Forest algorithm designed for analyzing human spatiotemporal trajectories in large-scale datasets. The algorithm leverages distributed computing frameworks to achieve efficient data partitioning and parallel processing, significantly reducing computational overhead. Advanced feature selection techniques, including Gini importance and Principal Component Analysis, were integrated to enhance the model’s ability to manage high-dimensional data while retaining critical predictive features. Experimental results demonstrate that the proposed algorithm achieves near-linear scalability, reducing training time by 7.5 times compared to traditional methods, while maintaining a high classification accuracy of 92.5% on complex datasets. The findings highlight the algorithm’s ability to uncover spatiotemporal mobility patterns, identify critical hotspots, and provide actionable insights for urban planning and resource optimization. This work addresses the computational limitations of traditional Random Forest implementations and establishes a robust framework for trajectory analysis in big data contexts. The proposed approach offers significant potential for applications in smart city development, real-time traffic management, and mobility behavior studies, providing a foundation for further exploration of adaptive and real-time data analysis techniques.