Answers are grounded in this paper and cite their sources.
This model reduces MSE by 2.7–41.3% in traffic forecasting, suggesting strong robustness and efficiency in Intelligent Transportation Systems.
Traffic time series forecasting is pivotal to Intelligent Transportation Systems (ITS), enabling the dynamic optimization of road networks. However,existing methods have limitations: traditional statistical models struggle to capture complex temporal patterns, while deep learning approaches, Transformer-based models are computationally intensive and impractical for real-time use. Additionally, many models ignore the distinct trend and seasonal components in traffic data, compromising forecasting accuracy. To overcome these challenges, this paper proposes Temporal Convolutional Linear Network (TC-Linear), a lightweight time series forecasting model that innovatively integrates the local perception of convolution with the global modeling capability. This combination enhances the computational efficiency for accurate traffic speed forecasting. Experimental results on large-scale datasets show that TC-Linear outperforms baselines in multi-variable long-term tasks, reducing MSE by 2.7–41.3% and MAE by 0.9–12.4%. Beyond enhancing the efficiency and interpretability of traffic time series forecasting, our model also exhibits strong robustness to noise, providing a reliable technological foundation for smart city decision-making systems.
No takes yet. Share an insight, caveat, or question.
Qin et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: