发表机构
Neapolis University Pafos; The Cyprus Institute; CaSToRC（帕福斯尼阿波利斯大学; 塞浦路斯研究所; CaSToRC）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出因果归因分数（CAS）架构，通过构建局部到全局的因果报告层，在基准和 DoubleML 数据集上验证其能有效区分预测结果与因果效应异质性的修饰因子，提升因果解释的准确性。
AI 中文摘要
预测性解释方法会对模型输出进行归因，但它们本身不会对现实世界结果的干预效应进行归因。我们提出因果归因分数（CAS），这是一种用于因果解释的紧凑分数架构。CAS 从已识别的干预联盟博弈出发，利用因果夏普利贡献分配联合干预对比，并将这些原始结果规模效应转换为局部 CAS、带符号的局部 CAS 以及两个互补的全局 CAS 摘要。创新之处不在于新的夏普利公式，而在于具有明确干预目标的局部到全局因果报告层。在已知真实值的基准测试中，八项重复的主要交互模拟（每次 n=2200，三种动作）显示，考虑联盟的 CAS 的局部 CAS 平均绝对误差（MAE）为 0.107，而单次归一化的 MAE 为 0.173，全局归一化绝对平均处理效应（ATE）向量的 MAE 为 0.213。与单次归一化相比，配对优势从可加性下的 -0.003 增加到强交互下的 0.091。在两个经验 DoubleML 数据集上，401(k) 资格/净资产（n=9915）和宾夕法尼亚州再就业奖金/失业持续时间（n=5099），预测性 SHAP/TreeSHAP 排名与因果效应修饰因子的 Feature-CAS 排名存在显著差异。在宾夕法尼亚州数据中，dep1（恰好一个受抚养人）从预测性全局排名 13 升至 Feature-CAS 排名 2，且是领先的局部 Feature-CAS 修饰因子。这些结果凸显了将预测结果的因素与解释估计因果效应异质性的因素相分离的附加价值。
英文摘要
Predictive explanation methods attribute a model output; they do not, by themselves, attribute an intervention effect on the real-world outcome. We introduce the Causal Attribution Score (CAS), a compact score architecture for causal explanation. CAS starts from an identified interventional coalition game, allocates the joint intervention contrast with causal Shapley contributions, and converts those raw outcome-scale effects into Local CAS, Signed Local CAS, and two complementary Global CAS summaries. The innovation is not a new Shapley formula, but a local-to-global causal reporting layer with an explicit intervention target. In the known-truth benchmark, eight repeated primary-interaction simulations (n = 2,200 each, three actions) gave mean Local CAS MAE of 0.107 for coalition-aware CAS, compared with 0.173 for one-at-a-time normalisation and 0.213 for a global normalised absolute ATE vector. The paired advantage over one-at-a-time normalisation increased from -0.003 under additivity to 0.091 under strong interactions. On both empirical DoubleML datasets, 401(k) eligibility/net financial assets (n = 9,915) and Pennsylvania reemployment bonus/unemployment duration (n = 5,099), predictive SHAP/TreeSHAP rankings differed materially from Feature-CAS rankings of treatment-effect modifiers. In Pennsylvania, dep1 (exactly one dependent) moved from predictive global rank 13 to Feature-CAS rank 2 and was the leading local Feature-CAS modifier. These results isolate the added value of separating what predicts the outcome from what explains heterogeneity in an estimated causal effect.
Comments10 pages, 5 figures
