Related Experiment Video
Updated: Aug 14, 2026
16:14
Trajectory Data Analyses for Pedestrian Space-time Activity Study
Published on: February 25, 2013
Multi-Perspective Spatio-Temporal Feature Fusion Model for Urban Traffic Flow Prediction
Avazjon Marakhimov1, Rustem Jalelov1, Jabbar Kudaybergenov2
1Department of Information Processing and Management Systems, Tashkent State Technical University, Tashkent 100174, Uzbekistan.
Sensors (Basel, Switzerland)
|August 13, 2026
Summary
Accurately forecasting urban traffic flow is challenging due to non-linear dynamics. The Multi-Perspective Spatio-Temporal Feature Fusion Model (MPSTFFM) enhances traffic prediction by integrating diverse spatial and temporal data views.
Area of Science:
- Urban planning and transportation science
- Artificial intelligence and machine learning
- Data science and network analysis
Background:
- Urban traffic flow prediction is complex due to non-linear dynamics and multi-range dependencies.
- Existing models struggle to capture the intricate spatio-temporal relationships inherent in traffic systems.
- Accurate traffic forecasting is crucial for efficient urban mobility and infrastructure management.
Purpose of the Study:
- To introduce a novel model, the Multi-Perspective Spatio-Temporal Feature Fusion Model (MPSTFFM), for accurate urban traffic flow forecasting.
- To effectively capture and integrate diverse spatial and temporal dependencies in traffic data.
- To improve the performance of traffic prediction systems by leveraging complementary data views.
Main Methods:
- Separating temporal signals into trend and fluctuation components.
- Representing spatial structures using four distinct graph types: first-order adjacency, second-order in-degree, second-order out-degree, and a data-adaptive graph.
- Employing self-attention for long-range dependencies and convolutional operations for local patterns within each view, followed by feature fusion.
Main Results:
- The MPSTFFM significantly outperforms twelve baseline methods on real-world traffic datasets.
- Achieved average reductions in Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) of 13.04%, 5.28%, and 9.59%, respectively.
- Demonstrated superior performance compared to existing state-of-the-art traffic forecasting techniques.
Conclusions:
- The MPSTFFM effectively models complex spatio-temporal dependencies for improved urban traffic flow prediction.
- The multi-perspective approach and feature fusion strategy are key to the model's enhanced predictive accuracy.
- This model offers a promising advancement for intelligent transportation systems and urban traffic management.