Hybrid Deep Learning for Spatiotemporal Traffic Forecasting: Integrating LSTM, Transformer, and Graph Convolutional Networks on the METR-LA Dataset
How to cite (IJASEIT) :
Accurate traffic prediction in large cities such as Los Angeles is increasingly necessary as cities expand and more vehicles are added to the roads. Using the METR-LA dataset. Using the METR-LA dataset, this study proposes a hybrid deep learning architecture that combines time and space modeling techniques to improve the accuracy and scalability of traffic flow predictions. The dataset consists of multivariate time series data from 207 loop detectors that record traffic speeds every five minutes with very high resolution. This study evaluates five potential model configurations: Long Short-Term Memory (LSTM), Transformer-based TSFormer, a combination of LSTM and TSFormer, Spatio-Temporal Graph Convolutional Network (STGCN), and a model combining STGCN and TSFormer. The evaluation conducted using three performance metrics Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) were used to assess how well each model captures complex temporal and spatial relationships. Our results show that the LSTM+TSFormer hybrid model consistently outperforms all other models across all criteria. This model has the lowest MAE (0.0624) and RMSE (0.1204), meaning it is better at learning patterns that occur over time and patterns that occur rapidly. STGCN-based models are quite good at capturing spatial dependencies, but their performance improves when combined with attention-based TSFormer modules. The hybrid models introduced in this study overcome major limitations, including the narrow receptive range of recurrent networks and the inflexible spatial structures assumed in graph-based methods. This work offers important perspectives for developing forecasting models that are not only accurate and scalable but also transparent and adaptable. Future work may explore dynamic graph construction and multimodal input integration to further enhance adaptability in real-world applications.
C. Lin and Z. Zha, “Los Angeles Metropolitan Traffic Forecasting using Graph Neural Network,” 2024 6th Int. Conf. Mach. Learn. Big Data Bus. Intell. MLBDBI 2024, pp. 139–142, 2024, doi:10.1109/MLBDBI63974.2024.10823861.
Q. Wu, Q. Fu, and M. Nie, “Graph Wavelet Long Short-Term Memory Neural Network: A Novel Spatial-Temporal Network for Traffic Prediction,” J. Phys. Conf. Ser., vol. 1549, no. 4, 2020, doi: 10.1088/1742-6596/1549/4/042070.
W. Zhang and K. Zheng, “Disentangled traffic forecasting via efficient graph neural network,” J. Phys. Conf. Ser., vol. 2858, no. 1, 2024, doi: 10.1088/1742-6596/2858/1/012036.
S. Wang, Y. Lv, Y. Peng, X. Piao, and Y. Zhang, “Metro Traffic Flow Prediction via Knowledge Graph and Spatiotemporal Graph Neural Network,” J. Adv. Transp., vol. 2022, 2022, doi: 10.1155/2022/2348375.
A. Oluwasanmi, M. U. Aftab, Z. Qin, M. S. Sarfraz, Y. Yu, and H. T. Rauf, “Multi-Head Spatiotemporal Attention Graph Convolutional Network for Traffic Prediction,” Sensors, vol. 23, no. 8, pp. 1–16, 2023, doi:10.3390/s23083836.
B. Lei, S. Huang, C. Ding, and M. Filipovska, “Efficient Traffic State Forecasting using Spatio-Temporal Network Dependencies: A Sparse Graph Neural Network Approach,” pp. 1–11, 2022, [Online]. Available: http://arxiv.org/abs/2211.03033
J. Lin, Z. Li, Z. Li, L. Bai, R. Zhao, and Z. Chen, “Dynamic Causal Graph Convolutional Network for Traffic Prediction,” Proc. 2023 IEEE 3rd Int. Conf. Inf. Technol. Big Data Artif. Intell. ICIBA 2023, pp. 229–233, 2023, doi: 10.1109/ICIBA56860.2023.10164956.
Q. Ma, W. Sun, J. Gao, P. Ma, and M. Shi, “Spatio-temporal adaptive graph convolutional networks for traffic flow forecasting,” IET Intell. Transp. Syst., vol. 17, no. 4, pp. 691–703, 2023, doi: 10.1049/itr2.12296.
Y. Shin and Y. Yoon, “PGCN: Progressive Graph Convolutional Networks for Spatial-Temporal Traffic Forecasting,” IEEE Trans. Intell. Transp. Syst., vol. 25, no. 7, pp. 7633–7644, 2024, doi:10.1109/TITS.2024.3349565.
P. Brimos, A. Karamanou, E. Kalampokis, and K. Tarabanis, “Graph Neural Networks and Open-Government Data to Forecast Traffic Flow,” Inf., vol. 14, no. 4, 2023, doi: 10.3390/info14040228.
W. Zhong, H. Meidani, and J. Macfarlane, “Attention-based Spatial-Temporal Graph Neural ODE for Traffic Prediction,” arXiv, 2023, [Online]. Available: http://arxiv.org/abs/2305.00985
X. Han et al., “Ollivier–Ricci Curvature Based Spatio-Temporal Graph Neural Networks for Traffic Flow Forecasting,” Symmetry (Basel)., vol. 15, no. 5, pp. 1–18, 2023, doi: 10.3390/sym15050995.
R. Jiang et al., “Spatio-Temporal Meta-Graph Learning for Traffic Forecasting,” Proc. 37th AAAI Conf. Artif. Intell. AAAI 2023, vol. 37, pp. 8078–8086, 2023, doi: 10.1609/aaai.v37i7.25976.
V. Singh, S. K. Sahana, and V. Bhattacharjee, “Integrated Spatio-Temporal Graph Neural Network for Traffic Forecasting,” Appl. Sci., vol. 14, no. 24, 2024, doi: 10.3390/app142411477.
Z. Shao et al., “Decoupled Dynamic Spatial-Temporal Graph Neural Network for Traffic Forecasting,” Proc. VLDB Endow., vol. 15, no. 11, pp. 2733–2746, 2022, doi: 10.14778/3551793.3551827.
L. Chang et al., “Multiscale Spatio-Temporal Aware Graph Recurrent Neural Network for Tra c Prediction,” 2024.
Z. Xie, W. Lv, S. Huang, Z. Lu, B. Du, and R. Huang, “Sequential Graph Neural Network for Urban Road Traffic Speed Prediction,” IEEE Access, vol. 8, pp. 63349–63358, 2020, doi: 10.1109/ACCESS.2019.2915364.
X. Cheng, Y. He, P. Zhang, and Y. Kang, “Traffic Flow Prediction Based on Information Aggregation and Comprehensive Temporal-Spatial Synchronous Graph Neural Network,” IEEE Access, vol. 11, pp. 47469–47479, 2023, doi: 10.1109/access.2023.3257221.
W. Kong, Z. Guo, and Y. Liu, “Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting,” Proc. AAAI Conf. Artif. Intell., vol. 38, no. 8, pp. 8627–8635, 2024, doi: 10.1609/aaai.v38i8.28707.
R. Caetano, J. M. Oliveira, and P. Ramos, “Transformer-Based Models for Probabilistic Time Series Forecasting with Explanatory Variables,” Mathematics, vol. 13, no. 5, pp. 1–29, 2025, doi: 10.3390/math13050814.
B.-A. Dai, N. Lyu, and Y. Miao, “FasterSTS: A Faster Spatio-Temporal Synchronous Graph Convolutional Networks for Traffic flow Forecasting,” pp. 1–13, 2025, [Online]. Available: http://arxiv.org/abs/2501.00756
C. Zheng et al., “Spatio-Temporal Joint Graph Convolutional Networks for Traffic Forecasting,” IEEE Trans. Knowl. Data Eng., vol. 36, no. 1, pp. 372–385, 2024, doi: 10.1109/TKDE.2023.3284156.
S. Y. Han, Q. W. Sun, Q. Zhao, R. Z. Han, and Y. H. Chen, “Traffic Forecasting Based on Integration of Adaptive Subgraph Reformulation and Spatio-Temporal Deep Learning Model,” Electron., vol. 11, no. 6, 2022, doi:10.3390/electronics11060861.
C. Zhao, R. Liu, B. Su, L. Zhao, Z. Han, and W. Zheng, “Traffic Flow Prediction with Attention Mechanism Based on TS-NAS,” Sustain., vol. 14, no. 19, pp. 1–12, 2022, doi: 10.3390/su141912232.
A. K. Takele and B. Villanyi, “Resource Aware Long Short-Term Memory Model (RALSTMM) Based On-Device Incremental Learning for Industrial Internet of Things,” IEEE Access, vol. 11, pp. 63107–63115, 2023, doi:10.1109/access.2023.3289076.
S. Liu and X. Wang, “An improved transformer based traffic flow prediction model,” Sci. Rep., vol. 15, no. 1, pp. 1–13, 2025, doi: 10.1038/s41598-025-92425-7.
J. Lin and Q. Ren, “Rethinking Spatio-Temporal Transformer for Traffic Prediction:Multi-level Multi-view Augmented Learning Framework,” 2024, [Online]. Available: http://arxiv.org/abs/2406.11921
X. Chen, A. Xiao, W. Pang, X. Jiang, Z. Ren, and M. Tang, “Confidential Spatio-Temporal Graph Convolutional Networks with Differential Privacy for Traffic Forecasting,” 2024 5th Int. Conf. Electron. Commun. Artif. Intell. ICECAI 2024, pp. 66–70, 2024, doi: 10.1109/ICECAI62591.2024.10674876.
H. Deng, “Traffic-Forecasting Model with Spatio-Temporal Kernel,” Electron., vol. 14, no. 7, 2025, doi:10.3390/electronics14071410.
Z. Guo, M. Lu, and J. Han, “Temporal Graph Attention Network for Spatio-Temporal Feature Extraction in Research Topic Trend Prediction,” Mathematics, vol. 13, no. 5, 2025, doi: 10.3390/math13050686.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Attribution-ShareAlike 4.0 International License
https://creativecommons.org/licenses/by-sa/4.0/