가로 보행환경은 주변 상업용 부동산 가격에 영향을 주는가? 해석 가능한 머신러닝과 딥러닝 기법의 적용

Despite the growing evidence demonstrating the importance of walkability for commercial property value, there is limited knowledge of which environmental characteristics contribute more or less to determining their values. Our study fills this gap by investigating the economic effects of walkable environments at different scales on commercial property values in Seoul, Korea. This study specifies walkable environments into macro-, meso-, and micro-scale and uses a computer vision technique to estimate streetscape features surrounding commercial properties. Based on these estimations, we employed various machine learning algorithms to produce optimal automated valuation models for the economic effects of walkable environments on commercial property values. Additionally, explainable artificial intelligence methods were used to identify variables that contributed the most to the black box models and to explore the non-linear relationships between walkable environments and commercial property values. Furthermore, our research demonstrated how the economic effects of walkable environments vary across submarkets classified by living population density. Our results indicate that various scales of walkability influence commercial property values, and these impacts differ according to submarkets. The findings of this study provide helpful insight into how to increase the economic benefits of commercial properties by creating walkable environments. Lastly, this study suggests practical approaches to explain the black box models with greater interpretability and transparency.