A Heterogeneous Ensemble Approach for Travel Time Prediction Using Hybridized Feature Spaces and Support Vector...

Area of Science:

• Intelligent Transportation Systems
• Machine Learning
• Deep Learning

Background:

• Accurate travel time prediction is vital for intelligent transportation systems, smart cities, and autonomous vehicles.
• Predicting traffic based on diverse factors is challenging.
• Ensemble learning approaches combining traditional and deep learning models show performance improvements.

Purpose of the Study:

• To propose an ensemble learning model for travel time prediction.
• To utilize hybridized feature spaces from deep learning models.
• To enhance prediction accuracy through a novel fusion technique.

Main Methods:

• Applied six state-of-the-art deep learning models (Bidirectional Long Short-Term Memory, Bidirectional Gated Recurrent Unit) to sensor traffic data.
• Fused feature spaces and decision scores from the best-performing model to create hybridized deep feature spaces.
• Employed Support Vector Regression on hybridized features for final travel time prediction.

Main Results:

• The proposed heterogeneous ensemble model demonstrated significant improvements over baseline techniques.
• Achieved a root mean square error of 53.87±3.50 and mean absolute error of 12.22±1.35.
• Obtained a coefficient of determination of 0.99784±0.00019, indicating high prediction accuracy.

Conclusions:

• The hybridized deep feature space concept yields more stable and superior travel time prediction results.
• The developed ensemble model offers a promising solution for accurate traffic forecasting.
• This research contributes to the advancement of intelligent transportation systems and smart city initiatives.