Related Experiment Video
Updated: Aug 19, 2025
16:14
Trajectory Data Analyses for Pedestrian Space-time Activity Study
Published on: February 25, 2013
13.6K
An attention-based recurrent learning model for short-term travel time prediction.
Jawad-Ur-Rehman Chughtai1,2, Irfan Ul Haq1,2, Muhammad Muneeb3
1Department of Computer and Information Sciences (DCIS), PIEAS, Islamabad, Pakistan.
Plos One
|December 1, 2022
Summary
This study introduces an attention-based Gated Recurrent Unit (GRU) model for improved short-term travel time prediction. The model enhances traffic forecasting accuracy by considering historical travel time sequence relationships.
Area of Science:
- Intelligent Transportation Systems (ITS)
- Data Science
- Machine Learning
Background:
- Intelligent Transportation Systems (ITS) are crucial for modern transportation, with travel time prediction (TTP) being a key component for congestion avoidance and route planning.
- Big Data and the Internet of Things enable advanced traffic analysis in smart cities using novel data sources like smartphones and navigation apps.
- Gated Recurrent Unit (GRU) models are effective for traffic prediction due to their ability to handle long sequences, but existing models overlook relationships within historical data.
Purpose of the Study:
- To develop an attention-based GRU model for enhanced short-term travel time prediction.
- To enable GRU models to learn relevant context from historical travel time sequences by updating hidden state weights.
- To improve the accuracy and reliability of traffic forecasting in intelligent transportation systems.
Main Methods:
- An attention-based Gated Recurrent Unit (GRU) model was proposed to capture dependencies within historical travel time data.
- The model was evaluated using Floating Car Data (FCD) from Beijing.
- Robustness was assessed through a noise-addition analysis using Gaussian distribution.
Main Results:
- The proposed attention-based GRU model demonstrated superior performance compared to existing deep learning time-series models.
- Key performance metrics including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R2) showed significant improvements.
- The model proved effective in accurately predicting short-term travel times.
Conclusions:
- The attention-based GRU model offers a significant advancement for short-term travel time prediction in intelligent transportation systems.
- The model's ability to consider contextual relationships in historical data leads to more accurate traffic forecasting.
- This approach provides a more reliable tool for urban traffic management and planning.
