This study aims to investigate and provide the most appropriate models estimating housing price. This study applies two approaches in modeling apartment prices in Seoul, Korea. First, existing methods such as Ordinary Least Square (OLS), Spatial regression, and multilevel models are applied. Second, multilevel models with spatial components are newly applied. In the process, explaining both sales and lease prices, this study applies various local features such as education, view, neighborhood markets, and medical services. The results show that the multilevel model with spatial component performs best. This study contributes to literature in two perspectives: 1) provision of new methodology in housing study by applying spatial component to multilevel modeling 2) provision of useful geographic information for housing and land management in Seoul.