Environmental, social, and behavioral determinants of residential stability in historic cities: evidence from Nara, Japan

Residential stability is a vital social indicator of community well-being and sustainability, yet its driving mechanisms remain insufficiently explored – particularly within the complex socio-spatial contexts of historic cities. These communities are shaped by the unique interplay of specific environmental conditions, social relationships, and behavioral dynamics, jointly influencing residents’ long-term attachment and settlement decisions. This study investigates 32 communities in central Nara, Japan, using data from 1,502 residents to develop an integrated analytical framework linking the built environment, social capital, and heritage usage behaviors. Employing Light Gradient Boosting Machine (LightGBM) and SHAP methods, the study identifies the most influential predictors of residential stability and uncovers their nonlinear and interactive effects. The results show that neighborhood interaction and geographical attachment strongly enhance residential stability, while excessive trust within tightly knit networks may weaken it. Heritage usage behaviors display dual effects: frequent but passive engagement tends to reduce stability, whereas active participation such as volunteering reinforces it. Built environment variables exhibit threshold and generational effects, and different heritage types show distinct impacts on stability. These findings highlight how environmental, social, and behavioral factors intertwine to shape residential patterns and population retention in historic cities, offering empirical insights into the spatial demography of community sustainability.