Skip to Main Content
 

Global Search Box

 
 
 
 

Files

ETD Abstract Container

Abstract Header

Benchmark, Explain, and Model Urban Commuting

Abstract Details

2012, Doctor of Philosophy, Ohio State University, Geography.
The research introduces a structured three-level framework -- system level, zonal level and interaction level of a commuting system to unify three main research fields in urban commuting studies. The system level deals with how to describe, evaluate, and compare urban commuting system(s) as a whole. The zonal level studies commuting spatial distribution and variation among commuting zones. The interaction level comes down to the building blocks of a commuting system – commuting flows -- with the goal of building models for them. The framework provides an all-around view and an integrated treatment towards urban commuting. At the system level, the random average commute is derived analytically based on the commuting matrix and the assumption of random pairing of origins and destinations. Commuting economy and normalized commuting economy metrics are adopted to evaluate commuting efficiency. At the zonal level, spatial autoregressive models are employed to provide robust explanations about the effects of a wide range of spatial and social dimension factors on the zonal variation of average out-commutes from origins. At the interaction level, a zero-inflated negative binomial model is introduced to address the over-dispersion issue of the Poisson models and the excessive zero-flow structure of commuting data at fine aggregation levels. It is a mixed model of a logit binary process and a negative binomial count process. Three dichotomous geographic relationship factors – intra-zonal indicator, neighboring indicator, and same county indicator, and the origin socio-demographic variables are incorporated into the spatial interaction model. The framework is applied to five U.S. metropolitan areas and, in greater detail, to Wichita, KS. It finds that the random average commute, defined as the average of a commuting distribution under the random assumption, coincides with the peak of the distribution. The two metrics adequately capture how far away the observed average commute departs from the random average commute. It also finds that spatial dimension factors, including locations, job accessibility and job-housing balance, have the most explanatory power on observed average out-commutes. Socio-demographic factors contribute to the model significantly after controlling for the spatial factors and spatial dependence. Results show that a zero-inflated negative binomial model greatly improves the model fit to the data, compared to the Poisson and negative binomial models. The three geographic relationship factors are found to contribute substantially to the model. Distance decay effects can be reinforced by these additional spatial relationship factors when zones vary dramatically in size. Quantitative explanation of factor influences on commuting flow generation is provided as an outcome of the statistical model.
Morton O'Kelly (Advisor)
Ed Malecki (Committee Member)
Ningchuan Xiao (Committee Member)
128 p.

Recommended Citations

Citations

  • Guo, M. (2012). Benchmark, Explain, and Model Urban Commuting [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1354597241

    APA Style (7th edition)

  • Guo, Meng. Benchmark, Explain, and Model Urban Commuting. 2012. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1354597241.

    MLA Style (8th edition)

  • Guo, Meng. "Benchmark, Explain, and Model Urban Commuting." Doctoral dissertation, Ohio State University, 2012. http://rave.ohiolink.edu/etdc/view?acc_num=osu1354597241

    Chicago Manual of Style (17th edition)