Spatiotemporal changes in behavior pattern of Nanjing's residents with unemployment-like activity-travel patterns

Investigating the spatiotemporal changes in the behavior of residents with unemployment-like activity-travel patterns (RUAT) elucidate how work-status-related disruptions are reflected in daily mobility. Existing research has primarily relied on interviews and questionnaires, which do not reveal spatiotemporal changes or driving mechanisms. Nanjing City, a super-large city in China, is considered in the analysis of the patterns and mechanisms of RUAT's activity-travel behavior. Nanjing comprises two main cities, three municipal cities, and nine new cities, reflecting varied levels of economic development. First, we used geographic big data to identify RUAT and construct a travel feature system. Second, we employed autoencoder and MiniBatchKMeans algorithms to categorize RUAT. Finally, we computed travel intensity and used multiscale geographically weighted regression to analyze the driving factors. The results indicated that the maximum trip distance, probability of visiting commercial areas, and average distance between stops decreased by ≥ 2.52% after changes in working status. Across the main, municipal, and new cities, RUAT with three different activity-travel patterns formed a concentric distribution, showing marked differences in features such as maximum trip distance and travel frequency. RUAT with the shortest average travel duration had a 72.08% lower travel intensity than those in city centers. Travel intensity changes in main and municipal cities were driven by built environment factors, whereas in Lishui and most new cities, sociodemographic factors played key roles. The results revealed geographic segregation in the driving mechanisms of spatiotemporal changes (built environment vs. sociodemographic factors), offering a new perspective for analyzing spatiotemporal behavioral changes.