Real Estate Price Evaluation Using Machine Learning Models

Accurate real estate price estimation is a critical task in property markets, influencing both buyers and sellers. Traditional valuation methods often struggle to capture the complex, nonlinear relationships among property features and market trends. This thesis explores the application of machine learning models for predicting apartment prices using a dataset containing real estate listings from a city in Kyrgyzstan. Six models are evaluated: Linear Regression, Ridge, Lasso, Decision Tree, Random Forest, and Support Vector Machine. Two target variables are analyzed — total price and price per square meter — allowing for comparison of direct versus normalized pricing approaches.Model performance is assessed using multiple evaluation metrics, including Mean Absolute Error, Root Mean Squared Error, R-squared, and Mean Absolute Percentage Error. Random Forest consistently outperforms other models, achieving the highest predictive accuracy and generalization across both targets. The results demonstrate the effectiveness of ensemble learning in real estate valuation tasks and suggest practical implications for deploying ML models in automated property appraisal systems.