Abstract
The accurate estimation of real estate value helps the development of real estate policies that can respond to the complexities and instability of the real estate market. Previously, statistical methods were used to estimate real estate value, but machine learning methods have gained popularity because their predictions are more accurate. In contrast to existing studies that use various machine learning methods to estimate the transactions or list prices of real estate properties without separating the building and land prices, this study estimates land price using a large amount of land-use information obtained from various land- and building-related datasets. The random forest and XGBoost methods were used to estimate 52,900 land prices in Seoul, South Korea, from January 2017 to December 2020. The models were also separately trained for different land uses and different time periods. Overall, the results revealed that XGBoost yields a higher prediction accuracy. Whereas the XGBoost models were more accurate on the 2020 data than on the 2017–2020 data when analyzing residential areas, the random forest models were more accurate on the 2017–2020 data than on the 2020 data. Further analysis will extend the prediction model to consider submarkets determined by price volatility and locality.
Highlights
Real estate has few market participants because of its high value [1]
For random forest (RF) models M1 and M2, the accuracy on the 2017–2020 sample was higher than it was on the 2020 sample, whereas the accuracy rates obtained by all XGBoost models on the 2020 sample were higher than those on the 2017–2020 sample
A larger sample might be expected to have better prediction capabilities, but differing results were obtained by the RF and XGBoost models
Summary
The quick and accurate estimation of real estate values resolves this instability in the real estate market to a certain degree and helps establish real estate policies [3]. For these reasons, attempts have been made to increase the quality of data in the public and private sectors to improve the estimation of real estate values, increase the efficiency and accuracy of valuation, and build an automated valuation model [4]. The results revealed that the machine learning models (ridge, LASSO, and elastic net) had better prediction capabilities than conventional regression models
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.