From Prediction to Action — How to Learn Optimal Policies From Data

If you know how to build predictive models, you can leverage this knowledge to learn optimal policies – rules that tell you the best way to act in various situations – directly from data.

Policy optimization problems are very common in the business world (e.g., arguably, every personalization problem is a policy optimization problem) and knowing how to solve them is a data science superpower.

The following series of blog posts aims to give you that superpower 🙂

• In Part 1, I motivate the need to learn optimal policies from data. Policy optimization covers a vast range of practical situations and I briefly describe examples from healthcare, churn prevention, target marketing and city government.
• In Part 2, I walk through how to create a dataset so that it is suited for policy optimization.
• In Part 3, I describe a simple (and, in my opinion, magical) way to use such a dataset to estimate the effectiveness of any policy.
• In Part 4, I show how to use such a dataset to find an optimal policy.

Happy learning!