Abstract
The Longitudinal Employer-Household Dynamics Origin-Destination Employment Statistics (LODES) are an important city planning resource in the USA. However, curating these statistics is resource-intensive, and their accuracy deteriorates when changes in population and urban structures lead to shifts in commuter patterns. Our study area is the San Francisco Bay area, and it has seen rapid population growth over the past years, which makes frequent updates to LODES or the availability of an appropriate substitute desirable. In this paper, we derive mobility flows from a set of over 40 million georeferenced tweets of the study area and compare them with LODES data. These tweets are publicly available and offer fine spatial and temporal resolution. Based on an exploratory analysis of the Twitter data, we pose research questions addressing different aspects of the integration of LODES and Twitter data. Furthermore, we develop methods for their comparative analysis on different spatial scales: at the county, census tract, census block, and individual street segment level. We thereby show that Twitter data can be used to approximate LODES on the county level and on the street segment level, but it also contains information about non-commuting-related regular travel. Leveraging Twitter’s high temporal resolution, we also show how factors like rush hour times and weekends impact mobility. We discuss the merits and shortcomings of the different methods for use in urban planning and close with directions for future research avenues.
Highlights
Historical land-use and development patterns, coupled with federal, state, and local policies have resulted in severe imbalances between jobs and housing in many sprawling metropolitan regions [1]
The difference is strong at the census tract level, where 42.7% of Twitter flows happen within a tract, whereas the same is true for only 3.4% of the Longitudinal Employer-Household Dynamics Origin-Destination Employment Statistics (LODES) flows
To address research question 1, to what degree do commuter flow patterns identified in geo-social network data (GSND) correlate with official LODES commuting data, we analyzed the correlations among the OD pairs of both data sources at multiple spatial scales
Summary
Historical land-use and development patterns, coupled with federal, state, and local policies have resulted in severe imbalances between jobs and housing in many sprawling metropolitan regions [1]. In most major American metropolitan areas, the results are apparent—rising housing costs, long commute times and bad traffic. The aim of transportation planning is to reduce the friction of distance, thereby increasing people’s mobility [2]. By focusing on work commutes rather than non-work trips, policymakers streamlined and simplified the complexities of travel behavior, research has consistently acknowledged their importance and influence [3,4]. Much of what has been written about the journey to work appears to be based on conventional/traditional attitudes where the journey is characterized as a trip that occurs from a permanent residential location to a permanent work location. Transportation planning and policy developed in the
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
