A Novel STCA-LightGBM-SVR Hybrid Model for Port Cargo Throughput Prediction

Abstract

Accurate port cargo throughput prediction is critical for global supply chain optimization yet remains challenging due to complex spatiotemporal dependencies and heterogeneous data integration. This study proposes a novel STCA-LightGBM-SVR (spatiotemporal crossattention mechanism) hybrid model integrating spatiotemporal cross-attention mechanisms with a dual-stage ensemble learning framework—specifically, a developed LightGBM-SVR ensemble where LightGBM handles feature selection followed by SVR for nonlinear fitting—to address these limitations. The parallel-designed attention module dynamically captures seasonal patterns (peak attention weight 0.18 ± 0.03) and inter-port correlations (attention coefficients 0.12–0.81), while the dual-stage ensemble combines LightGBM’s feature selection (F1 score = 0.89) with SVR’s nonlinear fitting capability (15.2% R² improvement). Experimental results on multi-port datasets (2010–2023) demonstrate superior performance over state-of-the-art algorithms such as Boosted DeepVAR and AutoGluon-Timeseries, with 4.82 million tons RMSE (15% reduction), 21.3% MAE decrease in long-term predictions, and 23.4% error reduction for extreme events, showcasing robust spatiotemporal modeling and practical applicability in port operations.

Received: 3 August 2025 | Revised: 16 October 2025 | Accepted: 4 January 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data available on request from the corresponding author upon reasonable request.

Author Contribution Statement

Tianwen Zhao: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Hanyu Xu: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization.