Author(s)

Chengze Shang ¹

Affiliation(s)

¹ La Salle College, Hong Kong, China

Corresponding Author

Chengze Shang

ABSTRACT

In order to optimize the effective utilization of road resources, reduce economic losses, and enhance the efficiency of traffic management, accurate prediction of road congestion is of paramount importance. This study focuses on the problem of traffic congestion prediction and thoroughly analyzes traffic congestion data from both temporal and spatial dimensions.  In the spatial dimension, violin plots are employed to analyze the spatial variations of traffic congestion, revealing significant differences in congestion levels among different road coordinates. Subsequently, this paper selects data from the 8 AM time frame and utilizes a scatter plot matrix to explore the correlations between traffic congestion values at different coordinate points.  In the temporal dimension, noticeable differences in traffic patterns between weekdays and non-working days are observed. Weekdays exhibit two distinct traffic peaks, whereas non-working days have a single, longer-lasting peak.  For the traffic congestion prediction, two algorithms, Lasso and Ridge regression, are employed, and the existing data is subjected to predictive analysis. To further enhance the performance of the models, this paper employs grid search to identify the optimal hyperparameters for the models.  The research findings demonstrate that both models yield highly accurate predictions, with minimal differences between them. Specifically, the Mean Squared Error (MSE) is 131.563, the Root Mean Squared Error (RMSE) is 11.470, and the Mean Absolute Error (MAE) is 8.044. These evaluation metrics validate the effectiveness and reliability of this research in the field of traffic congestion prediction, providing robust data support for future traffic management endeavours.

KEYWORDS

Traffic Congestion, Lasso Regression, Ridge Regression

CITE THIS PAPER

Chengze Shang, Research on traffic congestion prediction based on lasso and ridge regression. Frontiers in Traffic and Transportation Engineering (2024) Vol. 4: 70-78. DOI: http://dx.doi.org/10.23977/ftte.2024.040109.

REFERENCES

[1] Cheng Shuxiang, Xiong Shiqi, Liu Jun, et al. Highway Bridge Construction Cost Prediction Model Based on Ridge Regression Optimization Algorithm [J]. Construction Economics, 2023, 44 (S2):225-229.
[2] Niu Kai, et al. Prediction of Urban Traffic Congestion Level Based on BP Neural Network [J]. Tianjin Construction Science and Technology, 2021, 31(05):7-9.
[3] Tian Yu, et al. Traffic Congestion Prediction Model Based on Integrated Learning [J]. Computer & Telecommunication, 2020(04):60-63+70.
[4] Xiaoyu W, Xingyuan W, Bin M, et al. High-performance reversible data hiding based on ridge regression prediction algorithm [J]. Signal Processing, 2023, 204.
[5] Meihong S, Wenjian W. A network Lasso model for regression [J]. Communications in Statistics - Theory and Methods, 2023, 52(6).

Sponsors, Associates, and Links

---

• Advances in Naval Architecture and Ocean Engineering
  
  
• Journal of Vehicle and Intelligent Transport System
  
  
• Urban Transport Systems
  
  
• Journal of Connected Vehicles