Barrier Effect in a Medium-Sized Brazilian City: An Exploratory Analysis Using Decision Trees and Random Forests

This study aims to examine if an urban road with intense motorized traffic in a medium-sized Brazilian city constitutes a barrier for walking trips. A questionnaire was conducted with 103 individuals in an area up to 800 m from the road selected for the study to obtain information about personal characteristics (age, income, etc.), social interactions in the neighborhood, and travel and mobility characteristics. We used the dataset to explore the potential of Decision Tree and Random Forest classification models to predict the users’ perception of the barrier effect, which was characterized by the dependent variables speed and volume (of motorized traffic). For 36.9% and 47.6% of respondents, traffic speed and traffic volume, respectively, represent a barrier to walking. The results also show that the following variables considerably affect the perception of the barrier effect of the respondents: distance from their residence to the studied road, time living at the address and in the study area, social connections in the neighborhood, and the street reported as the busiest one in the neighborhood. Identifying the variables with the largest influence on the perception of the barrier effect may be very useful for planning and policy initiatives.