Sydney’s residential relocation landscape - Machine learning and feature selection methods - Unpack the whys and whens

Aug 4, 2026

Upload by Liam

This study investigates household residential relocation timing, a key input for transport and urban planning. Using a high-dimensional dataset of 1,024 relocations in Sydney, Australia, it compares ten machine learning survival techniques with three classical survival models. The study shows that when classical models are combined with tree-based automated feature selectors, they closely match machine learning results, with GBM, XGBoost, and Random Forest performing best. Feature analysis identifies homeownership as the leading predictor and highlights accessibility and recent life events as influential.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages32

Reading Time1 h 21 min

AccessFree to download

Introduction

Residential mobility decision framework

Focus on relocation duration

Role of transport-related attitudes

Key modeling topics: accessibility, daily trip travel-time, self-selection

Sydney’s residential relocation landscape: Machine learning and feature selection methods unpack the whys and whens Author: Bostanara , Maryam; Siripanich , Amarin; Ghasri , Milad; Rashidi , Taha Hossein Publication details: Journal of Transport and Land Use v. 17 Chapter No. 1 pp. 369-399 1938-7849 (ISSN) Publication Date: 2024-05-17 Publisher DOI: [https://doi.org/10.5198/jtlu.2024.2440](https://doi.org/10.5198/jtlu.2024.2440) License: [https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/) Link to license to see what you are allowed to do with this resource. Downloaded from [http://hdl.handle. net/1959.4/102225](http://hdl.handle. net/1959.4/102225) in [https://](https://)[ ](https://)[unsworks. unsw.edu.au](unsworks. unsw.edu.au) on 2024-10-11 Sydney’s residential relocation landscape: Machine learning and feature selection methods unpack the whys and whens Maryam Bostanara (corresponding author) rCITI, UNSW Sydney [m.bostanara@unsw.edu.au](m.bostanara@unsw.edu.au) Milad Ghasri UNSW Canberra [m.ghasri@unsw.edu.au](m.ghasri@unsw.edu.au) Amarin Siripanich rCITI, UNSW Sydney [a.siripanich@unsw.edu.au](a.siripanich@unsw.edu.au) Taha Hossein Rashidi rCITI, UNSW Sydney [rashidi@unsw.edu.au](rashidi@unsw.edu.au) Abstract: This study investigates household residential relocation timing, an aspect vital for transport and urban planning. Analyzing a high-dimensional dataset from 1,024 relocations in Sydney, Australia, the research contrasts ten machine learning survival techniques with three classical survival models. Results indicate that when classical models are paired with tree-based automated feature selectors, they align closely with machine learning outcomes. Notably, the GBM, XGBoost, and Random Forest models emerge as standout performers. The study provides a comprehensive comparison between automatic and manual feature selection, shedding light on variables influencing households ’ duration of stay. While stacked ensemble modeling, which leverages predictions from various models, is used to enhance accuracy, the improvements are marginal, underscoring inherent modeling challenges, particularly the recurring issue of misclassifying specific pairs of households in the concordance index measure. A thorough feature analysis highlights homeownership as the foremost predictor, underscoring the importance of recent life events and accessibility features in relocation decisions. The research emphasizes the importance of considering the accessibility of both current and future homes in relocation models, with 20% feature significance in model outcomes. Building on these foundational insights, the study paves the way for a deeper understanding of individual decision-making processes in sustainable urban planning. Keywords: Residential relocation, machine learning, survival analysis, residential self-selection, accessibility Article history: Received: October 17, 2023 Accepted: January 15, 2024 Available online: May 17 , 2024 1 Introduction The decision of residential mobility is one of the most significant choices households make multiple times throughout their lives. This decision typically involves several subdecisions, including the decision to leave the current home, choosing the new home ’s suburb and characteristics, determining the relocation timing, and more (Rashidi & Ghasri, 2017) . While these decisions occur at the household level, collectively they shape the housing market, neighborhood dynamics, city environment, and broader policies (Lerman, 1975) . Consequently, residential relocation has captivated researchers across various fields such as transport planning (Aditjandra et al., 2016), geography (Buckle, 2017), and economics (Sánchez & Andrews, 2011) for decades. From a transport planning viewpoint, understanding how households make residential decisions and how these relate to their transport attitudes and trip gen

FAQs

What is the document’s main research focus?

It focuses on household residential relocation duration and how timing relates to factors such as socio-demographics, financial status, homeownership, and life-course events.

How do machine learning survival methods compare with classical survival models?

The study contrasts ten machine learning survival techniques with three classical survival models and finds that classical models paired with tree-based automated feature selectors align closely with machine learning outcomes.

Which factors most strongly influence relocation timing and duration?

Homeownership is identified as the foremost predictor, and accessibility features along with recent life events are emphasized as important drivers of relocation decisions.

Aug 4, 2026

Upload by Liam