비수도권 아파트 가격 및 전세가 결정 요인 분석

광주, 대전, 대구, 부산의 사례를 중심으로

This study aims to analyse the determinant factor of apartments price and jeonse price in Gwangju, Daejeon, Daegu, Busan using various prediction models which consider spatial dependence among apartments price data. This study employs the dataset of apartment complexes price in Gwangju(716), Daejeon(426), Daegu(844), Busan(1,116) in 2017 for empirical comparison and 5 different prediction models such as Ordinary Least Square, Spatial Lag Model, Spatial Error Model, Geographically Weighted Regression, Geographically Generalized additive model. After empirical analysis, we come to the following conclusions. The larger the average area of building, the higher the total number of apartments, apartments those constructed by the construction company in the top 30, the better the educational environment, the higher the apartment prices. The closer to the convenience facilities such as parks, libraries, subways, banks and bus stops, the higher the apartment prices. The closer to social welfare facilities, inns, forests and cemeteries, the lower the apartment prices. However, the detailed results differed by city. We expect other research related housing price prediction will be considered both spatial dependence and non-linearity of a hedonic function to improve the performance of prediction.