发表机构
Walmart Global Tech（沃尔玛全球科技）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对电商复购预测的二分类模型栈，提出用生存模型替代，通过AFT模型及校准方法，在杂货电商数据上实现了更优的性能与效率，明确了校准与排名的权衡关系。
AI 中文摘要
电子商务中的复购推荐通常被建模为二分类问题，即“该客户会在W天内购买该商品吗”，该公式需要为每个感兴趣的时间窗口单独训练一个模型。我们用直接预测复购时间的生存模型替代该模型栈，并在某大型杂货电商平台的数百万客户数据上，对超过30种 ablation（消融）配置进行评估。本研究有三项贡献：第一，经验风险分析显示边际风险略有下降（k≈0.9），与“杂货商品自上次购买后间隔越久，复购概率越高（风险递增，k>1）”的普遍直觉不同。对数正态分布（Log-Normal）在边际拟合上表现最佳（R²=0.998），且排名性能最优，尽管威布尔分布（Weibull）在条件残差拟合上最佳，我们详细分析了这一明显差异。第二，单个加速失效时间（AFT）模型替代了三个按时间窗口划分的二分类器，在每个时间窗口的性能与后者相当或更优，同时总树数量减少了约3倍。在生存目标下，特征重要性发生了重新排序：渠道节奏和近期信号的重要性上升，而总频率计数的重要性下降。第三，一个4参数参数校准方法将原始生存累积分布函数（CDF）映射到各时间窗口的概率，且无跨时间窗口的单调性违反。AFT家族的校准质量相差一个数量级：指数AFT（Weibull k=1）的预期校准误差（ECE）约为1e-4，比对数正态分布低约10倍，而排名指标的相对差异在0.3%以内。我们将指数AFT用于需消耗概率的场景，对数正态分布用于纯排名场景，揭示了单一AFT家族内的原则性校准-排名权衡。
英文摘要
Repurchase recommenders in e-commerce are commonly framed as a binary question asking "will this customer buy this item within W days", a formulation that requires a separately trained model for every horizon of interest. We replace this stack with survival models that predict time-to-repurchase directly, and evaluate them on millions of customers from a major grocery e-commerce platform across more than thirty ablation configurations. Our study makes three contributions. First, an empirical hazard analysis reveals a slightly decreasing marginal hazard (k ~ 0.9), differing from the common intuition that grocery items become more likely to be repurchased the longer since the last purchase (increasing hazard, k > 1). Log-Normal achieves the best marginal fit (R^2 = 0.998) and the best ranking, despite Weibull providing the best conditional residual fit, revealing an apparent discrepancy we analyze in detail. Second, a single Accelerated Failure Time (AFT) model replaces three per-horizon binary classifiers, matching or exceeding each at its own horizon while using roughly 3x fewer total trees. Feature importance reshuffles under the survival objective: channel-cadence and recency signals rise while aggregate frequency counts fall. Third, a 4-parameter parametric calibration maps raw survival CDFs to per-horizon probabilities with zero cross-horizon monotonicity violations. Calibration quality varies by an order of magnitude across the AFT family: Exponential AFT (Weibull k=1) achieves expected calibration error (ECE) ~1e-4, roughly 10x lower than Log-Normal, while ranking metrics agree within 0.3% relative. We adopt Exponential AFT for probability-consuming surfaces and Log-Normal for pure ranking, exposing a principled calibration-ranking trade-off within a single AFT family.
CommentsAccepted at ReSys 2026 RecTemp Workshop
