发表机构
Goizueta Business School; College of Computing; Pratt School of Engineering; Emory University; Georgia Institute of Technology; Duke University（戈伊苏埃塔商学院; 计算机学院; 普拉特工程学院; 埃默里大学; 佐治亚理工学院; 杜克大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究基于KKBox数据集，通过九窗口充分性曲线发现手动续费用户早期行为在45-90天窗口内对流失预测提升显著，但曲线随实验设计（如移动目标、特征集）变化，强调窗口充分性声明需明确队列构建、目标定义和特征族。
AI 中文摘要
多少天的早期行为足以用于订阅流失预测？在公开的KKBox数据集中，流失的早期指标通常是某人合同状态的指标；然而，当查看大量流失的手动续费细分群体时，访问早期行为会显著提高对该特定细分群体的预测（120天时PR +0.10）。九窗口充分性曲线显示在45-90天区间内存在一个收益递减的拐点。然而，对三种队列/任务设计进行压力测试表明，该曲线对于被测试的设计是独特的；例如，在我们使用移动目标的测试中，曲线会反转，并且可能根据使用的特征集而偏移。因此，任何窗口充分性声明都应说明其队列构建、目标定义和特征族。所有证据来自一个音乐流媒体数据集；该机制应具有普遍性，但幅度可能不通用。
英文摘要
How many days of early behavior suffice for subscription churn prediction? In the public KKBox dataset, the early indicator of churn is typically an indicator of someone's contract status; however, when looking in the heavily churned manual-renewal segment, having access to early behavior creates a substantial increase in prediction for that specific segment (PR +0.10 at 120 days). A nine-window sufficiency curve shows diminishing returns; descriptive inflection detectors place a knee between 30 and 120 days depending on the metric and model class. Stress-testing over three cohort/task designs shows that this curve is singular to the design being tested; for example, in our test with a moving target, the curve inverts and can shift depending on the feature set used. Therefore, any window-sufficiency claim should state its cohort construction, target definition, and feature families. A survival-filtered replication on educational-attrition data (OULAD) confirms the rising-curve shape in one behavior-dominated setting (floor ROC 0.56 vs. 0.88), though signal composition inverts.
