A Study on Prediction of Housing Price Using Deep Learning

Abstract

The purpose of this study is to estimate housing prices using deep running. The simple RNN, LSTM, and GRU models, which are evaluated to be suitable for time series forecasting, are based on the time series data of apartment real price index, interest rate, household loan, building permit area and consumer price index. As a result of the empirical analysis, it is confirmed that the prediction power of the GRU model is superior to that of the learning data by evaluating the performance of forecasting power on apartment real price index based on the RMSE value. On the other hand, in the verification data, it is confirmed that the prediction power of the RNN model is excellent. Also, if the performance of the deep running model is evaluated with accuracy, the accuracy of the RNN model and the GRU model is the highest. As a result of this study, the government needs to build and develop a system that can predict and diagnose the housing market by using the deep learning technique that combines artificial neural network and big data to advance the housing market.