Comparative Analysis of Advanced Models for Predicting Housing Prices: A Review

Abstract

Understanding the determinants of housing price movements is an ongoing subject of debate. Estimating these determinants becomes a valuable tool for predicting price trends and mitigating the risks of market volatility. This article presents a systematic review analyzing studies that compare various machine learning (ML) tools with hedonic regression, aiming to assess whether real estate price predictions based on mathematical techniques and artificial intelligence enhance the accuracy of hedonic price models used for valuing residential properties. ML models (neural networks, decision trees, random forests, among others) provide high predictive capacity and greater explanatory power due to the better fit of their statistical measures. However, hedonic regression models, while less precise, are more robust, as they can identify the housing attributes that most influence price levels. These attributes include the property’s location, its internal features, and the distance from the property to city centers.