Short-term Traffic Flow Prediction Based on Multi-machine Learning

Abstract

Classical traffic flow prediction methods usually only analyze and make decisions based on the data of the predicted road itself, and often less consider the traffic flow correlation between different roads in the same area. According to the characteristics of traffic flow data in urban core area, This study builds a multi-dimensional data model of traffic flow for multiple related roads in the same area. Based on the data model, a traffic flow prediction algorithm based on multi machine learning competition strategy is proposed. The main idea of the algorithm is to reduce the dimension of multidimensional traffic flow data by time series clustering, and then perform concurrent training by introducing multiple multi-machine learning methods. The training results obtain the optimal classifier group through competition.