Prediction of new housing prices in Changsha urban area based on multiple machine learning algorithms - A comparative analysis

Aug 4, 2026

Upload by Ezra

The property market has faced substantial pressure in recent years, reducing turnover and raising financial risk for developers, making accurate housing price forecasting critical for both stakeholders and consumers. Many existing prediction approaches are difficult to apply because of heavy data requirements and diverse regional factors. Using February 2024 new building data from Changsha, the study compares multiple machine learning regression models, selecting the best option by evaluating and contrasting forecast errors. Stepwise, robust, Lasso, ridge, OLS, XGBoost, and random forest are tested.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages24

Reading Time1 h 0 min

AccessFree to download

Introduction

Property market context and importance of forecasting

Limitations of existing models across economic cycles and regions

Traditional and macro-driven approaches

Article Prediction of new housing prices in Changsha urban area based on multiple machine learning algorithms: A comparative analysis Yin Junjia1,2,*, Aidi Hizami Alias1,2, Nuzul Azam Haron1, Nabilah Abu Bakar1 1 Department of Civil Engineering, Faculty of Engineering, Universiti Putra Malaysia, Serdang 43300, Malaysia 2 UPM-Bentley BIM Advancement Lab, Universiti Putra Malaysia, Serdang 43300, Malaysia * Corresponding author: Yin Junjia, [gs64764@student.upm.edu.my](gs64764@student.upm.edu.my) CITATION | Junjia Y, AliasAH, Haron NA, Bakar NA. Prediction of new housing prices in Changsha urban area based on multiple machine learning algorithms: A comparative analysis. City Diversity. 2024; 5(1): 2742. [https://doi.org/10.54517/cd.v5i1.2742](https://doi.org/10.54517/cd.v5i1.2742)
ARTICLE INFO | | --- | | Received: 3 June 2024
Accepted: 11 July 2024
Available online: 5 August 2024
COPYRIGHT | Copyright © 2024 by author(s) . City Diversity is published by Asia Pacific Academy of Science Pte. Ltd. This work is licensed under the Creative Commons Attribution (CC BY) license. [https://creativecommons.org/licenses/](https://creativecommons.org/licenses/)[ ](https://creativecommons.org/licenses/)[by/4.0/](by/4.0/) Abstract: As China ’s pillar industry, the property market has suffered a considerable impact in recent years, with a decline in turnover and many developers at risk of bankruptcy. As oneof the most concerned factors for stakeholders, housing prices need to be predicted more objectively and accurately to minimize decision-making errors by developers and consumers. Many prediction models in recent years have been unfriendly to consumers due to technical difficulties, high data demand, and varying factors affecting house prices in different regions. A uniform model across the country cannot capture local differences accurately, so this study compares and analyses the fitting effects of multiple machine learning models using February 2024 new building data in Changsha as an example, aiming to provide consumers with a simple and practical reference for prediction methods. The modeling exploration applies several regression techniques based on machine learning algorithms, such as Stepwise regression, Robust regression, Lasso regression, Ridge regression, Ordinary Least Squares (OLS) regression, Extreme Gradient Boosted regression (XGBoost), and Random Forest (RF) regression. These algorithms are used to construct forecasting models, and the best-performing model is selected by conducting a comparative analysis of the forecasting errors obtained between these models. The research found that machine learning is a practical approach to property price prediction, with least squares regression and Lasso regression providing relatively more convincing results. Keywords: property market; lasso regression; ridge regression; extreme gradient boosted regression; robust regression; house price forecast; random forest; machine learning 1. Introduction The real estate industry is fundamental in the economic systems of many developing countries such as China and Malaysia [1], directly impacting gross domestic product (GDP) and employment. Through real estate development, urban infrastructure is improved, and residents ’ quality of life is enhanced. The development of it helps stimulate domestic demand and positively impacts related sectors such as construction materials, stock exchange, furniture, and home appliances. According to the latest data from the National Bureau of Statistics [2], as shown in Figure 1, China’s housing market has changed dramatically in terms of its share and size of GDP. Nowadays, the main policy direction of “houses are for living, not for speculation”has not changed, and the capital of real estate enterprises is facing a tight situation [3]. Many enterprises, such as Evergrande Real Estate, have begun to be exposed to a massive debt crisis. Such changes are being directly reflec

FAQs

Why is predicting housing prices considered important in this research?

Accurate forecasts help developers and consumers reduce decision-making errors. The study highlights that property-market instability affects both stakeholder outcomes and overall welfare.

Which machine learning regression models are compared?

The research evaluates stepwise regression, robust regression, Lasso regression, ridge regression, OLS regression, XGBoost, and random forest regression.

How does the study choose the best forecasting model?

It compares models by calculating and contrasting their forecasting errors on Changsha February 2024 new building data, then selects the best-performing one.

Aug 4, 2026

Upload by Ezra