Open AccessENHANCING THE ACCURACY OF MALAYSIAN HOUSE PRICE FORECASTING: A COMPARATIVE ANALYSIS ON THE FORECASTING PERFORMANCE BETWEEN THE HEDONIC PRICE MODEL AND ARTIFICIAL NEURAL NETWORK MODEL
Author(s) -
Nurul Fazira Sa’at,
Nurul Hana Adi Maimun,
Nurul Hazrina Idris
Publication year - 2021
Publication title -
planning malaysia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.232
H-Index - 7
eISSN - 1675-6215
pISSN - 0128-0945
DOI - 10.21837/pm.v19i17.1003
Subject(s) - multicollinearity , artificial neural network , real estate , heteroscedasticity , econometrics , computer science , house price , hedonic pricing , hedonic index , transaction data , valuation (finance) , artificial intelligence , machine learning , regression analysis , economics , operations research , database transaction , price index , mathematics , finance , programming language
The Hedonic Price Model (HPM), a prominent model used in real estate appraisal and economics, has been argued to be marred with nonlinearity, multicollinearity and heteroscedasticity problems that affect the accuracy of price predictions. An alternative method called Artificial Neural Network Model (ANN) was identified as capable of addressing the shortcomings of HPM and produces superior predictive performance. Hence, this study aims to evaluate the forecasting performance between HPM and ANN using Malaysian housing transaction data from the period between 2009 to 2018, sourced from the Valuation and Property Service Department, Johor Bahru. The models’ performance was evaluated and compared based on their statistical and predictive performance. Results showed that ANN outperformed HPM in both statistical and predictive performance. This study benefits the expansion of academic and practical knowledge in enhancing the accuracy of house price forecasting.