Exploring Bicycle Route Choice Behavior with Space Syntax Analysis
PIs: Ziqi Song and Anthony Chen (Utah State University)
Cycling provides an environmentally friendly alternative mode of transportation. It improves urban mobility, livability and public health, and it also helps with reducing traffic congestion and emissions. Although the mode share of bicycle accounts for a relatively small percentage of all trips taken in the United States, cycling is gaining popularity both as a recreational activity and a means of transportation.
To better serve and promote bicycle transportation, there is an acute need to understand the route choice behavior of cyclists. Unfortunately, there are only a limited number of studies investigating this issue. Moreover, space syntax theory is a promising tool for modeling traffic demand, but its applicability of modeling cyclists’ route choice behavior and the relationship with other bicycle-related attributes has not been explicitly studied with real-world data.
The linear regression analysis between space syntax measurements (including global integration and local integration) and collected bicycle volumes demonstrate that global integration is not statistically significant and can hardly explain actual bicycle volumes, while local integration is statistically significant and exhibits a good R-square value of 0.396. Moreover, the regression results indicate that local integration is positively related to bicycle volumes, which is as expected. Local integration provides significant explanatory power in modeling cyclists’ route choice.
Through incorporating other bicycle-related attributes, the linear regression model can provide better explanatory power. Five bicycle-related attributes are considered in this project, including segment bicycle level of service, motor vehicle volume, link pollution exposure, presence of a bicycle facility, and average slope of terrain. Based on the bicycle volume data collected in Salt Lake City, Utah, a series of regression models are tested. The model that includes local integration and all five bicycle-related attributes has the highest R-square value of 0.731, however, the model is neither statistically significant (at 95% level of significance) nor reasonable. The model that includes local integration and motor vehicle volumes as the explanatory variables exhibits higher R-square value (0.547) compared to the model that only considers local integration. Moreover, the model is both intuitively reasonable and statistically significant (at 95% level of significance). Other models tested are statistically insignificant.
This projects proposes a methodology to apply space syntax theory to modeling bicycle traffic. Travelers’ cognitive understanding of the network configuration, which plays an important role in their route choices, is explicitly analyzed and modeled using space syntax theory. Linear regression is used to analyze the correlation between bicycle volumes and space syntax measurements. To improve the explanatory power of the model, a number of bicycle-related attributes are considered through multiple regression analysis. A real-world case study is conducted in Salt Lake City, Utah, to demonstrate the proposed methodology. The results show that a space syntax measurement (i.e., local integration) can explain the bicycle volume distribution fairly well. By incorporating another bicycle-related attribute (i.e., motor vehicle volume), the model improves significantly in describing bicycle movement. Therefore, the combination of the space syntax measurement and other bicycle-related attributes can provide better explanatory power in modeling bicycle traffic.
The findings in this project have importation implications in bicycle facility assessment. Space syntax theory is demonstrated to be a useful tool in modeling cyclist route choice and can be used to guide the design of networks to accommodate bicycle travel more efficiently.