2019 Volume 32 Issue 2 Pages 53-64
This study reveals the house rent formation mechanism and introduces the unexpected relationship between the living environment and the market value of houses across Japan. We first developed the micro geodata that contains the distribution of all housing buildings across Japan. In addition, we built a database that can evaluate the house values throughout Japan by integration of the 82 kinds of rent formation factor and information on minimum guaranteed rent given by spatial unit of the postal code with the micro geodata. Moreover, by aggregating this database by postal code unit, clustering it by the k-means ++ method, and sparse modeling for each cluster, we developed a method to automatically reveal the rent formation factor which contributes positively or negatively to the minimum guaranteed rent for each cluster. As a result, it was shown that the determinant of the market value of the housing is not necessarily dominant whether the living environment is good or bad. In particular, it can be said that the rent formation factors on disaster risk should ring the alarm as a situation to be rectified because there were many areas that did not necessarily receive the impact of disaster risk other than flooding.