|
Performance Evaluation of Machine Learning Algorithms and the Hedonic Model in Property Valuation: Emphasizing Structural and Spatial Attributes
|
Majid Saebi *   , Mohammad Karimi |
|
|
|
Abstract: (13 Views) |
Accurate valuation of residential properties is crucial for decision-making among stakeholders, housing market experts, and urban managers. Given the multifaceted and complex nature of factors influencing property values, integrating machine learning techniques with spatial analysis can significantly enhance prediction accuracy. In this research, a hybrid dataset comprising structural characteristics and spatial attributes extracted within a GIS environment was utilized. Employing four machine learning algorithms Random Forest, Gradient Boosting Regressor, XGBoost, and CatBoost the process of feature vector extraction and optimization was conducted for property valuation. The proposed framework is designed around two valuation approaches: the first models the total property value (final price of the residential unit), while the second estimates the price per square meter. By normalizing for area, the second approach facilitates better comparability among properties. In both scenarios, the performance of the machine learning models was compared against the classical Hedonic model based on Ordinary Least Squares (OLS) regression. The results of the study, conducted in District 5 of Tehran over a two-month period, indicate that machine learning models delivered superior performance compared to the Hedonic model in both approaches. Specifically, the CatBoost algorithm demonstrated the highest efficiency, achieving a Root Mean Square Error (RMSE) of 0.30, an R-squared (R2) of 0.82, and a Mean Absolute Percentage Error (MAPE) of 10.04%; whereas the Hedonic model achieved, at best, an RMSE of 0.36 and a MAPE of 12.13%. These findings underscore the significant impact of integrating spatial analysis with machine learning techniques particularly the CatBoost algorithm in the valuation of residential properties.
|
|
| Keywords: Residential Property Valuation, Geospatial Information System, Machine Learning, Hedonic Model, Spatial Attributes |
|
|
|
Type of Study: Research |
Subject:
GIS
Received: 2026/01/8 | Accepted: 2026/06/26
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
