Loading...

Hierarchical Machine Learning Framework for Housing Price Prediction: Designing Intermediate Models for Key

Taghavi, Raziyeh | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58468 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Rezapour Niari, Maryam
  7. Abstract:
  8. Accurate housing price prediction has long been a strategic issue in economic market analysis and urban decision-making. Due to the structural complexity of the housing market—characterized by hidden variables, incomplete data, and nonlinear relationships among influencing factors—traditional models often fail to provide sufficient accuracy. This study aims to design and evaluate a hierarchical machine learning framework to improve housing price prediction under the non-ideal data conditions of Iran’s market. In the first stage, latent or incomplete macroeconomic variables such as daily exchange rate and gold price were estimated using intermediate models, including the SARIMA algorithm. The outputs of these models, along with structural housing attributes, were then fed into the final price prediction model based on XGBoost. The results indicated that the proposed hierarchical model outperformed flat models in terms of both accuracy and interpretability. The final model achieved a coefficient of determination (R²) of 95.1% and a relative Error (RE) of approximately 9.5%, demonstrating its strong capability in housing price prediction. Feature importance analysis revealed that building area, geographic location, exchange rate, and distance to urban services were the most influential factors affecting housing prices. Beyond improving predictive accuracy, the proposed framework is capable of extracting hidden knowledge from incomplete datasets and, due to its modular structure, can be easily adapted and updated under varying market conditions. These characteristics make the model applicable across different regions and suitable as both a foundation for intelligent pricing and property advisory systems in digital platforms, and a decision-support tool for policymakers and housing market stakeholders
  9. Keywords:
  10. Housing Pricing ; Housing Price ; Machine Learning ; Price Forecasting ; Incomplete Data ; Extreme Gradient Boosting (XGBoost) ; Seasonal Autoregressive Integrated Moving Average (SARIMA) ; Hierarchical Framework

 Digital Object List

 Bookmark

No TOC