Using Spectral Clustering for Urban Rail Station Classification

For the purpose of identifying the function and positioning of the urban rail stations and providing further guidance for design and construction ,a classification method using spectral clustering is established .On the basis of defi-ning the impact factors of urban rail station properties ,the data from Xi'an Metro Line 2 for the present and planned char-acteristics years are utilized to evaluate the effects of station classification .The K-means cluster algorithm and spectral cluster methods are employed ,including the unnormalized spectral clustering algorithm ,SM algorithm and NJW algo-rithm .The test results indicate that the NJW algorithm within the spectral clustering algorithm can properly classify the station according to the station properties under the influence of the urban rail network development and changes inland use and station properties .