Predicting Real Estate Price Using Stacking-Based Ensemble Learning

Abstract

As one of the leading researches focusing on modern economics, real estate industry not only affects people's well-being but also has a close relationship with the national economy and social stability. Nevertheless, there are numerous complex factors that influence real estate prices, which makes house price forecasting remain a classic and challenging problem in the field of data analysis. The development of data mining and machine learning has greatly facilitated the analysis and extraction of useful information from complex data sets and the building of models to make predictions. In this study, a stacking-based ensemble model is proposed to identify potential links between property prices and various factors so that the more accurate prediction of property prices can be made. Some base predictive models, including linear regression, support vector regression, ridge regression, least absolute shrinkage and selection operator, machine language programs, random forest regression, and gradient boosting regression are trained to individually predict the estate price in the experiment. Then, the stacking-based ensemble model is obtained by integrating competent base predictive models and optimized using Grid search. The experimental outcomes indicate that the proposed model is superior to base predictive models and can be more accurate in predicting house prices.