The Random Walk Value for Ranking Spatial Characteristics in Road Networks. Issue 4 (18th September 2014)

• Record Type: Journal Article
• Authors: Wei, Xuebin; Yao, Xiaobai A.
• Is Part Of: Geographical analysis. Volume 46:Issue 4(2014)
• Page Start: 411
• Page End: 434
• Publication Date: 2014-09-18
• Subjects: Geography -- Methodology -- Periodicals; Electronic journals; 910.00182
• DOI: 10.1111/gean.12064
• Languages: English
• ISSNs: 0016-7363

Abstract

<abstract abstract-type="main"> <title> <x xml:space="preserve">Abstract</x> </title> <p>This study proposes a new network index at both nodal and link levels to rank the spatial characteristics of individual network components. The objective is to create a network metric that captures socioeconomic characteristics in urban environments. Because this index is based on the random walk simulation modeling strategy, it is coined the random walk value (RWV). An algorithm and an associated software tool were developed to calculate the RWVs of network components. Compared with other popular network indices, the unique advantage of the RWV is that this index considers not only spatial structural or topological characteristics, but also physical characteristics of network components. Two case study cities, the Chinese city of Wuhan and the U.S. city of Atlanta, were chosen to test the utility of the RWV. These two case studies yield several findings. First, the RWV is highly consistent with some of the most widely used network measures, such as closeness and connectivity measures, which was evidenced by strong correlations between RWV and other networks. Second, the RWV has been proven to be a good indicator of spatial importance and a better predictor of socioeconomic variables in urban environments. The RWV outperforms all other network indices in terms of its correlations with important socioeconomic variables and its ability to predict some of them. Third, both case studies confirm that the RWV can be a good substitute for some important socioeconomic variables, such as population density and job density, in spatial modeling. This finding is significant for studies when population and job data are not available and for studies that attempt to predict future scenarios.</p> </abstract>