Implementing doubly robust estimation in R

Let’s consider a scenario consisting of a floral mega-corporation evaluating the influence of packaging on flower sales during the festive season. In this domain, details matter, and numerous variables interplay to drive sales – factors such as the weather, festival proximity, store capacity, and competitors’ pricing strategies.

In this example, we’ll consider two models: one predicting sales based on packaging and other factors, and another estimating the likelihood of a store choosing better packaging. If the sales prediction model is accurate, but the packaging choice model isn’t, DR estimation still provides reliable results. This is because the accurate sales prediction compensates for the inaccurate packaging choice estimate. Conversely, if the packaging choice model is accurate but the sales prediction model isn’t, DR estimation remains reliable because the accurate estimate of packaging...