Modeling Positional Uncertainty Acquired Through Street Geocoding

Modeling Positional Uncertainty Acquired Through Street Geocoding

Hyeongmo Koo (The University of Texas at Dallas, Richardson, USA), Yongwan Chun (The University of Texas at Dallas, Richardson, USA) and Daniel A. Griffith (The University of Texas at Dallas, Richardson, USA)
Copyright: © 2018 |Pages: 22
DOI: 10.4018/IJAGR.2018100101

Abstract

This article describes how modeling positional uncertainty helps to understand potential factors of uncertainty, and to identify impacts of uncertainty on spatial analysis results. However, modeling geocoding positional uncertainty still is limited in providing a comprehensive explanation about these impacts, and requires further investigation of potential factors to enhance understanding of uncertainty. Furthermore, spatial autocorrelation among geocoded points has been barely considered in this type of modeling, although the presence of spatial autocorrelation is recognized in the literature. The purpose of this article is to extend the discussion about modeling geocoding positional uncertainty by investigating potential factors with regression, whose model is appropriately specified to account for spatial autocorrelation. The analysis results for residential addresses in Volusia County, Florida reveal covariates that are significantly associated with uncertainty in geocoded points. In addition, these results confirm that spatial autocorrelation needs to be accounted for when modeling positional uncertainty.
Article Preview
Top

Introduction

Uncertainty is inevitably embedded in any process involving measuring and representing the world, and uncertainty in spatial data and spatial data analysis has been extensively investigated in Geographical Information Science (Griffith, Wong, & Chun, 2015). Positional uncertainty, which is one of five uncertainty types identified by the United States (U.S.) Federal Geographic Data Committee (FGDC), refers to inaccurate locations of geographic features (ANSI 1998). A common source of positional uncertainty is inaccuracy of the global positioning system (GPS) and geocoding errors. Horizontal and vertical positional accuracy of GPS points at reference stations have been reported regularly by the Federal Aviation Administration, suggesting that GPS recordings generally are reliable regardless of study area. In contrast, positional uncertainty of geocoded points can vary more, affected by geocoding algorithms and characteristics of study regions (Jacquez, 2012), as well as properties of street networks (Zimmerman & Li, 2010).

Understanding geocoding uncertainty is important to properly interpret spatial analysis results, which can be largely affected by the positional uncertainty of geocoded points (e.g., Burra, Jerrett, Burnett, & Anderson, 2002; Griffith, Millones, Vincent, Johnson, & Hunt, 2007; Harada & Shimada, 2006). Uncertainty can contribute to an increase in the standard errors of parameter estimates, and, subsequently, a reduction in the statistical power of a spatial cluster and/or trend detection (Lee, Chun, & Griffith, 2017; Zimmerman & Li, 2010). Previous studies (e.g., Bichler & Balchak, 2007; Cayo & Talbot, 2003; Hart & Zandbergen, 2013; Jones et al., 2014; Zandbergen, 2008a) emphasize that identifying geocoding errors can help researchers understand a geocoding error structure and set up an appropriate model specification for subsequent data analysis. These studies mainly focus on the magnitudes of geocoding uncertainty for various settings and locations. To achieve a reduction of uncertainty in a geocoding process and spatial analysis results, uncertainty modeling often is considered as a prerequisite process (Zhang & Goodchild, 2003), involving an exploration of possible covariates of the uncertainty. This investigation of possible covariates (i.e., properties of street networks) that potentially have an impact on the positional uncertainty of geocoded points can provide an insight into understanding geocoding positional errors. Rhwaw covariates can be utilized for identifying their impact on spatial analysis results (Zimmerman & Li, 2010), and for predicting positional uncertainties at specific locations (Jacquez, 2012). Interestingly, although Griffith et al. (2007) and Zimmerman, Li, & Fang (2010) show that the magnitude of geocoding positional uncertainty can be spatially autocorrelated, this data feature has received only limited attention in modeling positional errors of geocoded points (Zandbergen et al., 2012).

Complete Article List

Search this Journal:
Reset
Open Access Articles
Volume 12: 4 Issues (2021): Forthcoming, Available for Pre-Order
Volume 11: 4 Issues (2020): 3 Released, 1 Forthcoming
Volume 10: 4 Issues (2019)
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing