A data-driven predictive model for residential mobility in Australia – A generalised linear mixed model for repeated measured binary data. (September 2016)

• Record Type: Journal Article
• Authors: Namazi-Rad, Mohammad-Reza; Mokhtarian, Payam; Shukla, Nagesh; Munoz, Albert
• Is Part Of: Journal of choice modelling. Volume 20(2016)
• Page Start: 49
• Page End: 60
• Publication Date: 2016-09
• Subjects: Behavioural model -- HILDA data -- Household relocation modelling -- Mixed effects discrete-choice logit model (MEDCLM) -- Repeated measurements; Decision making -- Periodicals; Social choice -- Periodicals; Decision making; Social choice; Periodicals; 302.13
• DOI: 10.1016/j.jocm.2016.04.006
• Languages: English
• ISSNs: 1755-5345

Abstract

Abstract: Household relocation modelling is an integral part of the Government planning process as residential movements influence the demand for community facilities and services. This study will address the problem of modelling residential relocation choice by estimating a logit-link class model. The proposed model estimates the probability of an event which triggers household relocation. The attributes considered in this study are: requirement for bedrooms, employment status, income status, household characteristics, and tenure (i.e. duration living at the current location). Accurate prediction of household relocations for population units should rely on real world observations. In this study, a longitudinal survey data gathered in the Household, Income and Labour Dynamics in Australia (HILDA) program is used for modelling purposes. The HILDA dataset includes socio-demographic information such as general health situation and well-being, lifestyle changes, residential mobility, income and welfare dynamics, and labour market dynamics collected from the sampled individuals and households. The technique presented in this paper links possible changes in households' socio-demographic characteristics to the probability of residential relocation by developing a mixed effects discrete-choice logit model (MEDCLM) for longitudinal binary data using the HILDA dataset. The proposed model captures the effect of repeated measurements together with the area-specific random effects.