A Novel Deep Neural Network based Method for House Price Prediction

Abstract

Nowadays, house price is extremely essential for human living and plays an important part in the financial market, especially in the real estate industry. In this paper, we proposed a deep learning based network for house price prediction. Firstly, a brief introduction and related work are discussed. Then, Linear Regression, Random Forest Regression, XGBoost, and SVM are all detailed explained, developed and tested. In addition, several types of densely connected based neural network are proposed and developed. Finally, all methods are evaluated on a publicly available dataset, Boston house price dataset. Our method performs competitively performance compared with other classical methods including Linear Regression, Random Forest Regression, XGBoost, and SVM, and evaluated with a variety of evaluation methods including R2, Adjustable R2, MAE, MSE, RMSE.