Affinity Propagation Clustering in Defining Level of Service Criteria of Urban Streets

Abstract

Defining Level Of Service (LOS) criteria is very important as this is the first step of LOS analysis but this is not well defined in Indian context. The analysis followed in India is basically adaptation of Highway Capacity Manual (HCM 2000) methodology which is more suitable for developed countries having homogenous traffic pattern. An attempt has been made in this study to define LOS criteria of urban streets for developing countries having heterogeneous traffic flow condition. Defining LOS is basically a classification problem and to solve it Affinity Propagation (AP), a very recently developed cluster algorithm is used. Inventory details and the required speed data are collected from five major street corridors of Greater Mumbai Region in India through the application of Trimble GeoXT Global Positioning System (GPS) receiver. Six validation parameters are used on Free Flow Speed (FFS) data to find the optimal number of clusters, which is required for the classification of street segments into number of classes. After that speed data collected during both peak and off-peak hours are averaged over street segments and clustered into six groups to get the speed ranges of different LOS categories. Using validation parameters, considering the physical and surrounding environmental characteristics it is found that street segments can be classified into four classes in Indian context as mentioned in Highway Capacity Manual 2000. However, the FFS range for urban street class IV (urban design category) is significantly lower because of varying road geometric characteristics. The speed ranges of LOS categories under urban street classes are proportionately lower to that values mentioned in HCM 2000 because of highly heterogeneous traffic flow on urban Indian roads. The travel speed data collection procedure using GPS is simple and accurate. In addition, AP clustering is highly efficient in terms of time saving and provides a very accurate solution to classification problems. Hence, both GPS and AP techniques can be applied in other countries to define the speed ranges of LOS categories considering the local conditions.

Self Organizing Map of Artificial Neural Network for Defining Level of Service Criteria of Urban Streets

“…Bhuyan and Rao (2010) Cetiner et al (2010) developed a back propagation Neural Network traffic flow model for prediction of traffic volume of Istanbul City. The model uses the historical data at major junctions of the city for prediction of future traffic volume.…”