Research on the road traffic accident prediction based on SARIMA-LSTM model

Tong Cheng

doi:10.1117/12.3054553

20 December 2024 Research on the road traffic accident prediction based on SARIMA-LSTM model

Tong Cheng

Author Affiliations +

Proceedings Volume 13421, Eighth International Conference on Traffic Engineering and Transportation System (ICTETS 2024); 134213Z (2024) https://doi.org/10.1117/12.3054553
Event: Eighth International Conference on Traffic Engineering and Transportation System (ICTETS 2024), 2024, Dalian, China

Abstract

This study explores the necessity and importance of improving the accuracy of road traffic accident (RTAS) prediction in terms of human casualties caused by road traffic accidents globally. Taking China as an example, we analysed the increase in RTAS due to population and economic development in the last decade. By comparing the performance of three models, SARIMA, LSTM and SARIMA-LSTM, in RTAS prediction in Jilin Province, we find that the SARIMA-LSTM model significantly outperforms the SARIMA and LSTM models alone. For example, using 2019 data, the RMSE, MAE, and MAPE of the SARIMA-LSTM model improved by 51.79%, 50.84%, and 37.17%, compared with SARIMA, and by 27.51%, 21.85%, and 18.08%, compared with the LSTM model. This indicates that the hybrid model has higher prediction accuracy and adaptability when dealing with complex time series problems. This study provides a new perspective for RTAS prediction and demonstrates that combining traditional time series models with deep learning models can effectively improve the accuracy and robustness of prediction.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Tong Cheng "Research on the road traffic accident prediction based on SARIMA-LSTM model", Proc. SPIE 13421, Eighth International Conference on Traffic Engineering and Transportation System (ICTETS 2024), 134213Z (20 December 2024); https://doi.org/10.1117/12.3054553

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

;

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE