The registration data of real estate have high value because of its spatial attribute. Registration information mining can reveal the spatial feature of real estate registration. We explore the spatial model of real estate registration business of Wuhan with data mining in a big data perspective. First, we use global spatial anaylsis to reveal spatial autocorrelation for confirming spatial distribution pattern. Second, local spatial autocorrelation analysis is applied to reveal local distribution pattern, identifying spatial agglomeration, atypical rigion and outliers. Then, the K-Means clustering method is adopted to cluster the cadastral subareas. In order to comprehensively evaluate the clustering effect, we develop a new clustering evaluation index, proving that the real estate registration business has significant agglomeration characteristics. This article provides a path to reveal spatial feature of real estate registration for decision in land supply and property management of government.