Takahito Shimizu

The increase in vacant houses has become a serious social issue in many developed countries, including Japan. Therefore, to support mid- to long-term policy planning, there is a growing need to understand future vacancy distributions. In this study, we develop a machine learning model to predict future municipal-level vacancy rates by incorporating not only demographic and building information, but also spatial indicators related to road development conditions. First, we constructed a road mesh dataset at the 500-meter grid level by aggregating physical road indicators, block rectangularity, and the proportion of buildings with front roads. We then combined these variables with data from the Population Census and the Housing and Land Survey and developed a vacancy prediction model using LightGBM. The results show that incorporating road-related indicators improves prediction accuracy. In particular, we found that municipalities with a higher density of narrow roads, more irregularly shaped blocks, and a larger proportion of buildings lacking direct road access tend to have higher vacancy rates. This study demonstrates the value of road development information, which has received limited attention in previous research, in improving vacancy prediction, suggesting that road environments can influence the spatial distribution of vacant houses. Moreover, the findings of this study contribute to the early identification of areas at risk of high vacancy and the planning of preventive measures, thereby supporting urban management through the use of smart data.

Land use data, provided as open data by Japan’s National Land Numerical Information (NLNI), has long served as a fundamental resource across various fields such as urban planning and disaster prevention. The dataset divides the entire country into 100-meter square mesh units and classifies each unit according to its land use purpose, enabling spatially detailed understanding of land use patterns. However, maintaining this dataset requires significant time and cost, as human operators visually interpret land use for each mesh by overlaying satellite imagery with geospatial data. As a result, it is difficult to update the dataset rapidly on a nationwide scale, leading to insufficient responsiveness to changes in land use. To address this issue, this study developed a method to automate land use classification using satellite imagery and improve the efficiency of land use data maintenance. Specifically, the method utilizes high-resolution and high-frequency observation data from Sentinel-2 and employs machine learning to automatically classify land use into four categories: Residential land, Other inhabitable land, Water bodies, and Forests and wastelands. As a result, the method enables high-accuracy generation of land use data and achieves significantly improved efficiency compared to conventional approaches. Furthermore, by integrating time-series satellite imagery, the method shows potential for flexibly responding to changes in land use.

Understanding urban diversity is critical to inclusive planning for sustainable urban development. This study introduces a new Urban Diversity Index (UDI) based on global point-of-interest (POI) data for food-related establishments—defined here as facilities that offer food and beverage services, including various kinds of eating and drinking venues —covering 249 cities across 154 countries. The UDI integrates three components: Pielou’s Evenness Index (J′) to capture the balance of establishment types, a Coverage Ratio (C′) to measure global representativeness of establishment categories, and density (ρ′) to reflect spatial concentration. By applying concentric buffer analysis around city centers, we evaluate the spatial profiles of diversity in each city. Results show that while cities like London and Istanbul have similar index components, they exhibit significant differences in the spatial extent and pattern of high-diversity zones, reflecting their unique morphological and regulatory contexts. Furthermore, the analysis of “Peak Distance Buffer Zones”—areas where UDI remains above 95% of its maximum—reveals diverse urban forms, particularly in Asian megacities. Scatterplots of standardized UDI and peak distances identify distinct typologies of urban diversity structures. Notably, urban population size showed no significant correlation with UDI values. Overall, this study demonstrates the utility of globally standardized POI-based metrics in capturing the spatial heterogeneity of urban qualitative diversity and offers new insights into cross-city comparisons of urban complexity.