Authors: Devon Lechtenberg*, Capitol Region Council of Governments (CRCOG)
Topics: Transportation Geography, Spatial Analysis & Modeling, Urban and Regional Planning
Keywords: New England, Spatial Analysis, Commuting Flows
Session Type: Paper
Start / End Time: 2:00 PM / 3:40 PM
Room: Cleveland 1, Marriott, Mezzanine Level
Presentation File: No File Uploaded
Spatial autocorrelation in transportation network flows can be studied either by examining the attributes of their origins and destinations or by studying the attributes of the flows themselves. This research is an example of the former. Specifically, it concerns commuter flows originating in New England towns. The count of commuter outflows from a town is defined here as the number of destinations to which people living in a given town commute. The count of commuter outflows can serve as an indication of the relative accessibility of a town in relation to outside employment opportunities. Using methods of local spatial autocorrelation and geographically weighted regression (GWR), concentrations of towns with a higher count of commuter outflows were identified, followed by a GWR of the influence of commuting population, density of the highway network, and mean travel time on the count of commuter outflows. Globally, the coefficient of determination (R-Square) was very strong whereas the local coefficients of determination varied between displaying strong explanatory power in major urban centers throughout New England and weaker explanatory power in portions of northern New England. As the results are mixed, additional research is needed in order to further refine the model specification and ensure proper spatial randomness is present in the GWR residuals. Nevertheless, it is argued that geographically weighted regression can enhance our understanding of factors which predispose a town or city to being accessible to many places of work as well as the study of transportation flows in general.