Real Time Vehicle Flow Prediction Based On Random Forest Algorithm

Abstract

Various applications include traffic forecasting, vehicle navigation and car pathway and congestion management, to gather accurate information on potential traffic transfers across the transportation network. A major constraint in getting traffic flow information in real time is that most connections are not connected to traffic sensors. Some variables, such as impacts, public events and shutdowns, that affect reliable traffic flows are often unforeseeable, and show that traffic forecasting is a challenge. First, we propose to use a dynamic simulator in all links using existing traffic, traffic history and expected request information. The prediction of traffic information and the state will bring important travel information to the attention of people. This paper also discusses the problem of prediction of traffic. A Random Forest approach is mainly used to foresee and analyze transport countries that consider traffic prediction as an issue in classification and not to forecast and analysis the parameters of traffic flow. Condition of traffic can be classified into six types by service level. Spatial and temporal properties which is not found in previous methods, can also be used together.