基于Bagging的交通拥堵预测研究

Study on Congestion Prediction Based on Bagging Arithmetic

针对交通拥堵原因的多元性及单个神经网络拥堵模型准确率不高的特点,设计了一个以BP神经网络为弱学习算法、基于Bagging集成学习方法的交通拥堵预测模型.与单个神经网络模型相比,Bagging后的预测模型具有更加优良的性能,可为市内交通预警决策提供分析与支持.

Congestion prediction is studied. Congestion prediction needs consider traffic factors and surrounding effects. At present, some researchers have developed the traffic prediction models using single neural network. But it is difficult to improve accuracy and stability of model based on single network. This paper combines many factors and presents a congestion prediction system which based on Bagging. The BP neural network is selected as the weak leaner. Experiments show that the system based on Bagging gets better per- formance than single neural network, and can serve as traffic early warning of decision-making.