Street Patterns Distinction Based on Logit Model

Abstract:

In this study, 718 traffic analysis zones were classified into 5 groups according to the features of the network. Then, 6 quantitative indexes of different road networks were calculated. It was concluded that Meshedness, proportion of cul de sacs and proportion of four leg intersection were the best measure to distinguish and describe various street patterns. At last, the multinomial logit model was developed based on the above mentioned indices to quantitatively distinguish street patterns, and the accuracy of the model was proved to be 88.4%, which was 3.2% higher than that of the visual inspection. This paper offers an approach to quantitatively distinguish street patterns, which can be used to study the relationship between street patterns and traffic performance.