Abstract. In this study, we analyze the quality of water hydrant data for estimating housing vacancies based on their spatial relationships with the other geographical data that we consider are correlated with such vacancies. We compare with in-situ vacant house data in several small districts, thus verifying the applicability of the water hydrant data to the detection of vacant houses. Through applying Bayesian approach, we apply the water hydrant data and other geographical data to repeatedly Bayesian updating for the classification of vacant / no vacant houses. We discuss the results of this classification using the temporal intervals associated with turning off metering, fluctuations in local population density, the densities of water hydrants as indicators of vacancies and several other geographical data. We also conduct the feasibility study on visualisation for the estimation results of housing vacancy distributions derived from the fine spatial resolution data.
Read full abstract