One important task of urban public transit network planning is to identify poorly served areas and improve their service level. Due to the limitation of analysis tools, planners have used a single index, such as transit network density, to evaluate the whole network service. Some serious deficiencies come with this single index evaluation: (1) poorly served areas can not be efficiently found by this method; and (2) the relationship of public transit service demand and supply has not been fully considered. To overcome those deficiencies, a GIS-based Fuzzy c-Means clustering analysis method has been put forward in this paper. Fuzzy c-Means clustering analysis, which has been widely used in pattern recognition, is a useful tool in classifying and understanding complicated systems. With GIS's powerful functions of topology analysis, public transit network indexes at traffic zone level can be calculated, and it becomes possible to penetrate the network and study differences between partitions of the network. Important indexes that identify the traffic zone's transit network service, eg transit network density, and population density, have been chosen as the input vectors of the Fuzzy evaluation system. Application of this GIS-based Fuzzy c-Means clustering analysis method in the Nanjing Public Transit Planning Project has shown good results. Poorly served areas were identified efficiently. The analysis results have been found to be correct by experts from local transit enterprises. With its Fuzzy characteristic, this method could be used in other cities without much adaptation. (a).
Samenvatting