The prediction of traffic flow with regression analysis

Abstract

Traffic data mining applying machine learning algorithms is necessary to analyse and understand the road traffic flow in busy cities. Also, it is very essential for making smart cities. Mining traffic data helps us to reduce travel delays and improve the city life. Currently, many cities in the developed countries use different sensors to collect the real-time traffic data and apply machine learning algorithms on the traffic data to improve the traffic condition. In this paper, we have collected the real-time traffic data from the city of Porto, Portugal, and applied five regression models: Linear Regression, Sequential Minimal Optimisation (SMO) Regression, Multilayer Perceptron, M5P model tree and Random Forest to predict/forecast the traffic flow of Porto city. Also, we have tested the performance of these regression models. The experimental results show that the M5P regression tree outperforms the other regression models.