Defining Pedestrian Level of Service at Signalized Intersections Through Modelled Parameters and Som Clustering

Abstract

Aim of this study is to define pedestrian levels of service (PLOS) categories (A-F) at signalized intersections under the influence of mixed traffic flow. Both quantitative and qualitative factors affecting the service measure have been identified and modelled using ridge regression method. Along with road geometric and traffic operational data, effective 675 pedestrian perceptions of real-time sense of satisfaction have been used in the model development. A PLOS score model has been developed taking perceived satisfaction score as the dependent variable and factors such as no of lanes, 85 th percentile speed of vehicles, volume of through moving vehicles, left turning motorized and non-motorized vehicles, permissible right turning motorized and non-motorized vehicles and pedestrian delay as independent variables. Delays used in PLOS model were estimated by the combination of waiting time delay and vehicle interaction delay. Having R 2 value of 0.986 PLOS model was well validated. PLOS scores of six categories were defined by applying Self-Organizing Map (SOM) in Artificial Neural Network (ANN) clustering. Analysis shows that about 75% of total crosswalks were offering average or above service levels, whereas only 4 % were contributing very poor kind of service to pedestrians in the study areas.