Distributed identification of heterogeneous treatment effects

In many areas including precise medical treatments and financial investments, analysis of heterogeneous treatment effects has become important. In this paper, we focus on identifying subgroups by combining data in a distributed storage system. We propose a distributed algorithm based on the alternating direction method of multipliers, which can well preserve privacy of subjects. This method can deal with large-scale data, and perform well in identifying subgroups if there exist sufficient samples in a whole distributed storage system but not necessarily in every computing node. Our numerical study indicates that the proposed method is promising in many interesting cases.