Spatio-Temporal Traffic Flow Prediction in Madrid: An Application of Residual Convolutional Neural Networks

Daniel Vélez-Serrano,Jose Vélez-Serrano,Alejandro Álvaro-Meca,Fernando Sebastián-Huerta

doi:10.3390/math9091068

Abstract

Due to the need to predict traffic congestion during the morning or evening rush hours in large cities, a model that is capable of predicting traffic flow in the short term is needed. This model would enable transport authorities to better manage the situation during peak hours and would allow users to choose the best routes for reaching their destinations. The aim of this study was to perform a short-term prediction of traffic flow in Madrid, using different types of neural network architectures with a focus on convolutional residual neural networks, and it compared them with a classical time series analysis. The proposed convolutional residual neural network is superior in all of the metrics studied, and the predictions are adapted to various situations, such as holidays or possible sensor failures.

Highlights

Union, with 3.1 million inhabitants and approximately 1.7 million vehicles circulating through its streets
The autoregressive integrated moving average (ARIMA) model corresponding to Equation (1) obtained an mean absolute error (MAE) of 40.85
From our point of view, our model provides two contributions to spatio-temporal problems: We showed that deep learning provides an important improvement over linear models, such as ARIMA or SARIMA, and neural networks were shown to be a very dynamic research area where there is still much room for improvement compared with the more classical methods

Summary

Introduction

With 3.1 million inhabitants and approximately 1.7 million vehicles circulating through its streets. A traffic forecasting system is essential for managing the growing volume of vehicles in these cities. In this respect, traffic flow prediction has received special attention in the last two decades [1]. Using spatio-temporal data [2] obtained from a range of sensors, a variety of real-world problems can be solved, such as the demand for taxis [3,4,5], urban traffic control and congestion avoidance [6,7], abnormal event detection [8,9], and travel time estimation or route planning [10,11,12], amongst others.

Objectives

Methods

Results

Discussion

Conclusion