The uneven spatial distribution of key urban resources, particularly employment opportunities, has profound implications for social equity and sustainable urban development. Using Urumqi as a case study, this research develops a framework based on multi-source big data that integrates mobile phone signaling records, residential community information, points of interest (POI), online recruitment postings and enterprise registries. Building on an interpretable CatBoost+SHAP machine learning approach, the study systematically evaluates inequalities between residential areas with different rent levels in terms of potential employment opportunities and residents’ actual employment opportunities. The results show that low-rent residential areas perform systematically worse than other residential areas in terms of potential employment opportunities, potential employment average salary and potential employment accessibility. Residents of these areas also experience structural disadvantages in securing actual employment opportunities, actual employment average salary and acceptable actual commute time. Additional analysis identifies two critical thresholds: monthly rents of approximately 20 CNY/m²/month and 23 CNY/m²/month mark significant turning points for potential employment opportunities and actual employment opportunities, respectively. These thresholds indicate a pronounced non-linear relationship and diminishing marginal returns between housing costs and employment access. By emphasizing the interaction between urban spatial structure and individual capabilities, this study clarifies the mechanisms through which jobs-housing mismatch translates into unequal opportunity access, and provides empirical evidence to support future research and assessment of employment disadvantages along the housing cost gradient.
Fan et al. (Fri,) studied this question.