This paper focuses on probabilistic risk assessment of detached houses in Hino city of Tokyo, and analyses the probabilities that a detached house would be abandoned with respect to the categorical data of geographical and demographical condition. Hino city has been developed along Tamagawa and Asakawa rivers and on Tama hills. Therefore, it is appropriate area for our research focusing on how geographical and demographical condition would influence on the risk of occurrence of abandoned house. We use not only statistical survey data but field survey data, which revealed incorrectly counted houses as abandoned, since we thought that there must be unknown factors influencing the statistically observed data.
From the analysis of this paper, we found that: 1) it was confirmed that the influence of the number of miscounted data in statistically collected data by applying Poisson regression analysis of the generalized linear model with integrated nested Laplacian approximation (INLA) using the number of abandoned houses (statistical number) and the questionnaire survey data, 2) the risk of occurrence tends to increase in the categories where the altitude difference is large. Its risk increases by nearly twice, 3) in the population categories, the risk of occurrence of the second and third categories appears high, there is a risk of becoming abandoned house where existing property and newly built property mix, 4) from the ratio of 65 years old or over, the risk of occurrence appears high in the district where the percentage of age group from 65 years old to 70 years old is high.
With the aging of the population and the declining population, abandoned house countermeasures have been studied and implemented nationwide. As administrative response, we are struggling with how to treat the house itself, upon request of the residents. One effective means, however, is to take measures before becoming an abandoned. For that purpose, we propose to collect more detailed and individually surveyed data.
In the Poisson regression analysis of the generalized linear model used in this paper, geographical and population requirements were selected as occurrence risk factors based on the location condition of Hino City, which is the analysis target, and the results of the past research. Of course, this is not the case in any city, and the factors selected depending on the location will change. However, analytical methods can be applied, and policy making based on risks is required. In the future, abandoned houses and unoccupied rooms countermeasures for apartment houses that were not covered in this paper are also urgent issues and I would like to make them future research subjects.
