计算机科学
操作化
透视图(图形)
体验式学习
服务(商务)
用户体验设计
逻辑回归
损耗
相互依存
潜在Dirichlet分配
结果(博弈论)
人气
多项式logistic回归
知识管理
社会技术系统
放弃(法律)
人机交互
竞赛(生物学)
主题模型
数据科学
过程管理
平滑的
期望理论
客户体验
再开发
营销
摘要
ABSTRACT The rapid growth of mobile e‐commerce has intensified competition for user retention, making customer churn a critical concern for digital platforms. User reviews contain rich signals about experience failures, yet existing approaches struggle to link specific UX frictions to churn outcomes. This research introduces a data‐driven framework that combines text classification, topic modeling and regression analysis to quantify how distinct friction themes influence churn intent. Churn intent is operationalized as explicit abandonment cues extracted from reviews, identified in 1.3% of all feedback and 13.4% of negative reviews. Of the 19 625 reviews retained after topic assignment, 494 carry explicit churn signals and serve as the dependent variable in all logistic models. Using BERTopic, 20 latent UX friction themes are derived, and logistic regression estimates their marginal effects on churn probability. The findings offer managers a prioritization model for addressing high‐impact UX issues. By treating churn intent as an emergent outcome of interacting experiential and operational subsystems, the study contributes a socio‐technical perspective on user attrition in digital service ecosystems.
科研通智能强力驱动
Strongly Powered by AbleSci AI