Product Analytics: Applied Data Science Techniques for Actionable Consumer Insights

13. Uplift Modeling

In Chapter 12, we covered the topic of statistical matching in an effort to extrapolate causal relationships from observational data. In this chapter, we’ll pivot to a whole different topic: We’ll focus on situations in which we have A/B testing results and want to further understand the individual treatment effects. Individual treatment effects are the causal effect of a particular treatment on an individual, rather than an aggregate group.

Recall that in the introduction to causal inference in Chapter 10, we discussed a notable limitation of such inference: We are estimating the group-level treatment effect. However, in reality, treatment effects are individual. An individual will see some effect of the treatment, and it ...