Fast Food Data: Where User‐Generated Content Works and Where It Does Not

Abstract

As big urban data usage expands in the social sciences, there remain real concerns about fidelity to on the ground conditions. In this paper, we examine the correspondence between Phoenix metro area restaurants identified by a social media source (http://yelp.com) and those from an administrative source (Maricopa Association of Governments [MAG]). We find that they capture largely disjoint subsets of Phoenix restaurants, with only about one‐third of restaurants in each data set present in the other. Point pattern analyses indicate that the Yelp data is significantly clustered relative to the MAG data. Specifically, restaurants in Yelp are concentrated in certain parts of metro Phoenix, most notably the downtowns of Phoenix, Scottsdale, and Tempe. Further analysis indicates that areas with more Yelp than MAG restaurants tend to have more college‐educated workers and workers employed in the Arts, Entertainment, and Recreation sector. Our comparison highlights the strengths and weaknesses of each data source: Yelp data is far more detailed and comprehensive in certain locations, while MAG data is more consistent across the entire region due to its systematic construction. When combined, administrative and user generated databases seem to provide a more holistic and comprehensive picture of the world than either would provide by itself.

Quantitative Geography III: Future Challenges & Challenging Futures

“…This data is not perfect (Wyly 2014;Ash et al 2018), but can be useful with the right theoretical grounding (Shelton et al 2015, 199). Indeed, bias from non-representative samples (Boeing and Waddell 2017;Folch et al 2018;Zhang and Zhu 2018) or spatial disparity in coverage (Shelton et al 2014) is increasingly well studied.…”

The US Census Longitudinal Employer-Household Dynamics Datasets

“…One aspect of this research includes studying vulnerability to emergencies (Kermanshah, Derrible 2016). The Workplace Area Characteristics files have been used to describe neighborhood character (Folch et al 2017) and to identify business districts (Manduca 2018). The LODES data have also been used for web visualization, including the "Where Are The Jobs?"…”

Crowdsourced Mapping for Healthy Food Accessibility in Dallas, Texas: A Feasibility Study

“…Another study, that classified restaurants by taste, found associations between restaurants labeled with certain tastes and nearby neighborhood characteristics, such as income, racial composition, and education (23). However, only a couple of studies have used Yelp data in healthy food access research (24,25). These studies demonstrated that Yelp data contained more accurate details on restaurants and healthy food stores than traditional business information sources and commercially available datasets, but had coverage limitations over a metro area.…”