Bicycle level of service in urban Indian context

Abstract

The purpose of this study is to define Bicycle level of service (BLOS) for urban Indian context based on the mixed traffic flow condition and various road attributes. Video camera was employed to collect the data sets from two cities, Rourkela and Bhubaneswar of Odisha State, India. The collected data are used to calibrate the BLOS model developed by the Florida Department of Transportation to find the BLOS score of each road segment. Calibrated model coefficients appropriate in Indian context are determined using multivariate regression analysis. In order to define levels of service provided by urban on-street segments, BLOS scores are classified into six categories (A-F) using k-mean, HAC, Fuzzy c-means, Affinity Propagation (AP), Self Organizing Map (SOM) and GA-fuzzy clustering methods. These clustering methods show different BLOS ranges for service categories. However, to know the most appropriate clustering technique applicable in Indian context, Average Silhouette Width (ASW) is calculated for every clustering method. After a thorough investigation it is induced that K-mean clustering method is the most appropriate one to define BLOS categories. The defined BLOS score ranges in this study are observed to be higher than that witnessed by FDOT studies; implies the kind of service the bicyclist perceived in urban Indian context is inferior to that observed by FDOT. From all the factors that affect BLOS score, “effective width of outside through lane” affects the most. The study concludes that bicyclist travel, more often, at the poor quality of service of “D”, “E” and “F”, than good quality of service of “A”, “B” and “C”. This may be due to lack of proper attention by the planners and developers towards bicycle facilities in urban Indian context.