Loading...

A Copula Based Joint Model of Residential and Work Location Choice

Movaghar Hoor, Sara | 2015

661 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47224 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Samimi, Amir
  7. Abstract:
  8. Residential location choice is one of the most important decisions that households should make. This decision has many effects on household's travel patterns and urban built environment. Considering predetermined residential locations is one of the weaknesses of these models. Joint modelling of residential and work location choices is a solution to this problem. So far, multinomial logit, nested logit and mixed logit are the structures that have been used for this joint modeling. Independence from irrevelant alternatives, forcing sequential structure and complicated calculations in large dimension problems are respectively weaknesses of mentioned structures. In this researsh a copula based joint model of residential and work location choice is represented. Since copula functions facilitate considering dependence effects by generating closed forms for probability expressions, application of this approach in econometrics has been increased in recent years.
    Data used in this research is from comprehensive transportation studies of Shiraz city in 1387. Inner part of the city has been divided in to 156 traffic zones, so for both decisions each person has 156 alternatives.In addition to independent models of residential and work location choices, joint model of two decision was built using several bivariate copula functions such as Frank, Gumbel, Clayton, and Joe. Several important results were obtainded from these models: First of all, the estimated copula parameters were positive and significant which implies there are correlated unobserved factors in residential and work location utility functions that increase (decrease) both utilities in the same direction. Second, the model built using Clayton function has the best fit to data. The Bayesian Information Criterion for this model and independent model is 17022 and 25162, respectively. Third, calculating percent correcltly predicted for models showed that joint models are not better than independent models in independentally predicting residential and work location, however, they have relatively better performance in joint prediction of these decisions. Percent correctly predicted is 0.15% for independent models and 0.38% for the joint Clayton model
  9. Keywords:
  10. Discrete Choice Model ; Copula Functions ; Shiraz ; Residential Location Choice ; Dependence Parameter

 Digital Object List

 Bookmark

No TOC