Application Of Data Mining Technology On Short-time Traffic Flow Proediction

With the progress of urbanization, and the development of the automotive industry, traffic has become a serious problem in today's society. Intelligent transportation system is an internationally recognized settlement of the best way to solve city and highway traffic congestion, improve operating efficiency, and reduce air pollution. And the accurate short-time traffic flow prediction is the premise and key of the intelligent transportation system to traffic flow controlling and inducing. Mainly studies on the prediction of short-time traffic flow using the data mining technology of a major traffic flow parameters (including traffic volume, average speed and lane share) based on road traffic flow controlling and inducing.In order to improve the accuracy of the prediction, before the data mining, first, data should be pre-processed including repairing, dispersing, noting, and plotting. Then proposed a method to fix data based on association rule mining algorithm and the experiments proved the efficiency of the implementation of GSP algorithm is better than Apriori algorithm. Then, high efficiency algorithm PrefixSpan is chosen for sequential pattern mining based on pre-processed traffic flow data, the frequent sequences are used to predict the next moment traffic flow situation. PrefixSpan algorithm is not suitable for multi-moment short-time traffic flow prediction, so in this paper PrefixSpan algorithm is improved, and used in multi-moment short-time traffic flow prediction. Experiments show that the algorithm played a good prediction results. It is beneficial for traffic control and traffic flow induction, improved the intelligent transportation system.