Stacking-Based Ensemble Learning Method for House Price Prediction

• Published in: Software Engineering Application in Informatics Vol. 232; pp. 224 - 237
• Main Authors: Liu, Yuanning, Wu, Yifan, Su, Linlin, Li, Wenxuan, Lei, Jianjun
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2021; Springer International Publishing
• Series: Lecture Notes in Networks and Systems
• Subjects: House price prediction; Machine learning; Regression model; Stacking ensemble model
• ISBN: 9783030903176; 3030903176
• ISSN: 2367-3370; 2367-3389
• DOI: 10.1007/978-3-030-90318-3_22

It is difficult for the empirical prediction to provide accurate prediction results for house price due to its frequent fluctuation. Motivated by recent developments and advantages of emerging machine learning algorithms, this paper proposes a stacking model to improve the prediction accuracy of house price, which merges several outstanding base models, Bagging regression, Extra-Trees regression, XGBoost and LightGBM. Meanwhile, we analyze all the factors affecting house price and present a more practical and complex data preprocessing method to select the most contributing features by combining a creative feature engineering method. Moreover, we also use linear model as the meta model. Experimental results reveal that the proposed stacking model has high prediction accuracy and obtains the best performance over all base models and the meta model especially in predicting extreme values.