A Study on Prediction of Seoul Prime Office Price Using Deep Learning

Abstract

1. CONTENTS (1) RESEARCH OBJECTIVES The purpose of this study is to establish an office price prediction model in a situation where Seoul prime office is preferred as an investment product for real estate funds and REITs in the domestic real estate indirect investment market. (2) RESEARCH METHOD The methodology of this study analyzed the VAR model, SimpleRNN model, and LSTM model using macroeconomic indicators and market condition indicators to compare office price prediction power. (3) RESEARCH FINDINGS As a result of the analysis, the RMSE of the VAR model was lower than that of the LSTM model and the RNN model, so it was analyzed that the predictive power was high. These analysis results empirically show that the VAR model, a multivariate time series eep learning model, an artificial neural network algorithm. 2. RESULTS As a result, this study confirmed that the Seoul prime office market operates in a linear relationship rather than a non-linear one between variables. In other words, compared to other real estate markets, the Seoul prime office market can be seen as a market in which prices are formed while exhibiting a stable relationship with interrelationships between variables rather than rapidly changing environmental factors such as macroeconomic variables in terms of price prediction through time series methodology.