A Housing Price Prediction Method Based on Stacking Ensemble Learning Optimization Method

Abstract

The growth of real estate sector has been significantly influenced in recent years by ongoing regulation of real estate acquisition policy and the effects of the COVID-19 epidemic on economy. The fluctuation in housing prices is one of the most concerning factors for prospective homeowners. Whether property prices can sustain a comparatively constant level for an extended period of time is a crucial factor for prospective homeowners. Numerous modeling and application techniques for prediction algorithms, together with the promotion of machine learning algorithms, offer fresh approaches to forecast residential real estate values. This work proposes a novel stacking ensemble learning method (DStacking) which is based on the diversity of learners including XGBoost and BP neural network. Through the application of ensemble learning algorithm, D-Stacking method can successfully predict the possible promotion of housing price. Housing price datasets from China and the USA were used in the experiments to guarantee the generalizability of findings. Experimental findings indicate that the diversity of base learners significantly affects the predictive power of D-Stacking method. Furthermore, the more diverse the models are, the more precise the predictions can be with proposed D-Stacking method. Compared with classical Stacking ensemble learning models, the proposed D-Stacking method demonstrates an excellent feasibility in reducing the RMSE to 0.869 and 1.029 across various datasets.