DEFINING LOS CRITERIA OF URBAN STREETS USING GPS DATA: K-MEANS AND K-MEDOID CLUSTERING IN INDIAN CONTEXT

Abstract

The objective of this study is to classify urban streets into a number of classes and to define the speed ranges of levels of service (LOS) categories in Indian context. In this purpose, average travel speed on street segments is used as the measure of effectiveness, which has been obtained from second-wise speed data collected using Global Positioning System (GPS) receiver. Midsized vehicle (car) was used to collect travel speed data on five urban road corridors comprising of 100 street segments in the city of Mumbai and two major road corridors of Kolkata city in India. Both k-means and k-medoid clustering methods and several cluster validation measures have been employed in the classification of urban streets and LOS categories. It is found that k-medoid clustering is more suitable in Indian context and speed ranges of level of service categories are significantly different from that values mentioned in HCM 2000.