자본 체급과 경쟁 밀도 기반 카페 생존 분석 및 폐업 예측 모형 — 서울특별시를 중심으로

초록

Purpose This study aims to identify the multidimensional determinants of café survival in the Seoul metropolitan area and to construct an early warning system for business closure. Methods A total of 39,898 business license records from January 2010 to November 2025 were collected from the Local Government Licensing Database and spatially merged with commercial district analysis data, yielding 28,590 final observations. Kaplan–Meier survival analysis, Cox proportional hazards modeling, Gaussian mixture model–based commercial district typology, and XGBoost classification were employed. Results Contrary to classical location theories by Hotelling (1929) and Christaller (1933), brand capital and competition density dominated closure risk over commercial district location. Commercial district presence increased closure hazard by 6% (HR = 1.06, p < 0.005). The Nelson–Aalen hazard rate identified 18–24 months post-opening as the critical risk interval. XGBoost achieved an ROC-AUC of 0.681, with brand type variables accounting for about 65% of total feature importance. Conclusion A Multi-layered Early Warning System (MEWS) combining structural risk scores with temporal weighting and macroeconomic adjustment was proposed to provide data-driven risk diagnostics for prospective entrepreneurs.

키워드