EconML

Problem Setup

We assume we have data that are generated from some collection policy. In particular, we assume that we have data of the form: , where is the observed outcome for the chosen treatment, is the treatment, are the co-variates used for heterogeneity, are other observable co-variates that we believe are affecting the potential outcome and potentially also the treatment ; and are variables that affect the treatment but do not directly affect the potential outcome. We will refer to variables as controls and variables as instruments. The variables can also be thought of as control variables, but they are special in the sense that they are a subset of the controls with respect to which we want to measure treatment effect heterogeneity. We will refer to them as features.

在EconML包的定义中，是指对个体的处理，则是在处理下的观测结果。是针对异质性的控制变量，是指可能会同时影响到处理和潜在结果的观测变量，是指会影响到处理但不会直接影响到潜在结果的变量。在这里，即为控制变量，而为工具变量。和均为控制变量，但主要为能够反映个体异质性的特征集合，它未必会影响处理，而更倾向于指能够影响到处理的控制变量。