The Application Of Clustering Analysis In Traffic Flow Time Series Data Mining

With the rapid development of ITS in China, it seems to be a problem that how to use the mass data to provide the meaningful forecast outcomes better contributing to the traffic management and Data mining technology serves as a powerful tool to solve it.Time series is one of the most common data type in Intelligent Transportation System, so the analysis of traffic flow time series turns particularly important. Utilizing clustering mining technology to analyze time series of traffic flow not only can discover the traffic flow rules behind mass data, but also can produce an optimal TOD (the time of day) plan. Furthermore, combined with some spatial information, some useful spatial and temporal distributed laws in transportation could be revealed. This paper presents a scheme of data driven methodology based hierarchical clustering that determines optimal break points for TOD traffic control after the first clustering process and proves the scheme's validation under C++ Builder and Matlab program. The study results show that the method produces fairly good clusters. In this paper, some research work on traffic flow time series data mining has been done. Firstly, clean the data and give compensation to the dataset. Secondly, propose a scheme based hierarchical clustering for the variation tendency of time series. Thirdly, analyze the clustering statistics and traffic flow time series clustering outcomes. Moreover, visualization of clustering results would be presented.