Machine Learning Models for Predicting Second-hand House Prices: A Comparative Study

This paper presents a comparative analysis of the application of machine learning models in predicting the prices of second-hand houses. It underscores the importance of the real estate market within the economy and points out the advantages of precise predictions for various stakeholders, including homebuyers, real estate agents, and financial institutions. Traditional methods frequently fail to grasp the intricate interactions that impact house prices, whereas machine learning techniques have garnered attention for their capability to handle such complexities. The study made use of the “Shanghai Second-hand Housing Prices from 2024 Anjuke Website” dataset from Kaggle, which comprises 175,135 records, each having nearly 30 features. After undergoing preprocessing procedures like duplicate removal, feature transformation, outlier removal, missing value imputation, feature encoding, feature engineering, and data standardization, the data was divided into training and testing sets. Four machine learning models were compared: Linear Regression, Ridge Regression, Random Forest, and XGBoost. The findings revealed that Random Forest and XGBoost surpassed the linear models, with Random Forest attaining the lowest Mean Absolute Error (MAE) and the highest R² Score. The study concluded that Random Forest was the most effective model for predicting the prices of second-hand houses in Shanghai. Even though the results were promising, the study acknowledged certain limitations such as the temporal and geographical constraints of the dataset and the reliance on historical data. Future research ought to concentrate on real-time data integration, model updating, incorporating additional data sources, and exploring advanced techniques to improve the performance and interpretability of the model.