Application of Machine Learning for Condominium Price Prediction Using Real-Time Web Scraped Data: Evidence from Hanoi’s Emerging Market

Real estate valuation and prediction in emerging markets often rely heavily on subjective judgment. The study examined current valuation practices at three real estate agencies in Hanoi. The findings show inefficiencies and biases in current practices. To address these challenges, this research aims to develop a data-driven framework that integrates real-time web-scraped data to predict condominium prices in Hanoi, Vietnam. The data set was scraped from the most popular proptech platform, batdongsan.com, from 18/08/2025 to 21/11/2025. The scraped data contains 22,565 observations and 16 explanatory variables. The research proposed deploying three ensemble learning models – Random Forest, XGBoost, and CatBoost – as suitable methods for the data conditions. The performance of these models was evaluated using RMSE, MAE, MSE, and R2. The research highlighted the need for modernization and transparency in property valuation in Vietnam and advocated data-driven practices.