Smart urban mobility is one of the key components to build future cities. From traffic congestion to increased travel times and pollution, most of the cities around the world are losing billions of money every year due to inefficient transportation planning. This work addresses these issues, by proving solution methodologies for travel time prediction to achieve efficient traffic management. With the improvements in processing power, RAM, data storage, etc., computationally expensive techniques such as deep learning are making their way into prediction. In this paper, we present two travel time prediction techniques, long short-term memory (LSTM) and random forest (RF). A real-life travel time data set for an urban road network of New South Wales is considered to perform numerical experiments. The performance of these techniques is analyzed and compared against some of the existing supervised learning techniques. Various prediction horizons have been investigated to validate these models. It is evident from the simulation results that the LSTM and RF models demonstrate promising predictions with very low prediction errors, varying between 0.5% to 6.8%.
Data-driven Models for Short-term Travel Time Predictio
01.10.2019
380674 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Short-term travel time prediction
Online Contents | 2003
|Short-term Travel-time Prediction on Highway: A Review of the Data-driven Approach
Online Contents | 2015
|Short-term Travel-time Prediction on Highway: A Review of the Data-driven Approach
Taylor & Francis Verlag | 2015
|Short-Term Prediction of Highway Travel Time Using Multiple Data Sources
Springer Verlag | 2015
|Short-Term Travel Time Forecasting in Hong Kong
ASCE | 2004
|