Application Of Machine Learning In Traffic Flow Prediction

Machine learning is a cross subject in a number of areas such as probability theory,statistics,operations research and approximation theory,and is widely used in the field of computer vision,medical diagnosis,biometric identification technology and so on.Traffic flow prediction is a hot spot in the research of intelligent transportation system,which is of great significance for relieving urban traffic pressure and reducing environmental pollution.In this article,we first introduce development history of machine learning and traffic flow prediction theory,then introduce boosting tree,time series analysis,neural network,and present some solutions to deal with lots of missing values,shortage of samples and making good use of topological information.And we compare boosting tree model,time series analysis model and neural network model's model performance in public data sets and analyze their advantages and disadvantages.The results show that boosting tree model is better than other two models in traffic flow prediction.Considering the complex nature of traffic flow,we combine bagging algorithm with boosting tree to reduce prediction variance.The results show that this method can improve model performance significantly.