Micro Level Speed Choice Behavior on a Rural Highway in an Heterogeneous Traffic Environment: Latent Class Modeling Approach

Abstract

The modeling of driver speed choice behavior on a rural highway with heterogeneous traffic environment in a developing country is reported. The study explores the application of a latent class model as an alternative methodological approach, to relate factors such as road geometry, roadside environment, traffic mix and operational characteristics with speed choice at the micro level. In the conventional analysis based on lane-based homogeneous traffic environments, the impact of exogenous variables is kept constant across road segments; whereas those impacts might vary across different segments for different conditions. To check this possible variation, the study compares latent class linear regression with ordinary least square, as well as the random parameter linear regression model. The proposed models were estimated using data collected from two sections of a major national highway in Bangladesh. The entire roadway section was divided into several small segments and the critical segment length was estimated based on several sensitivity tests. The small segmental level exogenous variables were considered for the model building and estimation. The results obtained could be used to understand better the speed choice factors of drivers in the non-lane-based traffic environments of developing countries. Law enforcement agencies, traffic operation and maintenance officials, as well as roadway design and planning authorities, could gainfully use the findings of this study to inform more responsive speed management strategies.