我们提出了一个强化学习框架可以巧妙的运用在结合AB实验的结果,帮助企业优化消费者在浏览,点击,购物的转化路径上每一个节点的触达。
Song, Yicheng, and Tianshu Sun. "Ensembling Experiments to Optimize Interventions along Customer Journey: A Reinforcement Learning Approach." Management Science, Forthcoming.
我们利用机器学习估计个体的处理效应异质性(heterogeneous treatment effect)结合混合整数规划帮助企业和政策制定者最大化不同政策带来的因果效应。
McFowland III, Edward, Sandeep Gangarapu, Ravi Bapna, and Tianshu Sun. "A prescriptive analytics framework for optimal policy deployment using heterogeneous treatment effects." MIS Quarterly 45, no. 4 (2021).
如何在缺少个人数据的情况下应用已有AB实验的结果帮助决策?我们提出了一个鲁棒优化框架,帮助企业和政策制定者在缺失个人数据的情况下做最鲁棒的决策。
Gupta, Vishal, Brian Rongqing Han, Song-Hee Kim, and Hyung Paek. "Maximizing intervention effectiveness." Management Science 66, no. 12 (2020): 5576-5598.