Support vector machine and entropy based decision support system for property valuation

Abstract

Property valuation is crucial to real estate developers, financial institutions and buyers as it could help determine the financial viability, establish a fair value of a real estate scheme, and eliminate the risk of borrowing respectively. Advanced mathematical algorithms such as artificial neural network (ANN) and support vector machine (SVM) may open up new ways to improve the valuation accuracy. This research aims to present an overview of the potential suitability of the SVM technique for property valuation. It is proceeded by identifying the key variables which could affect the property price. An entropy‐based rating and weighting method has been presented with the aim of providing objective and reasonable weighting. Then, based on the key variables, the predictive ability of SVM is compared with multiple regression analysis (MRA) and ANN outcomes. The results obtained from practical case studies in Hong Kong and mainland China indicate that, entropy and SVM serve better function for factor weighting and property valuation respectively. Hence, an entropy and SVM based decision support system is proposed, in which the key variable selection and the price valuation are integrated.