From Meaningful Data Science to Impactful Decisions: The Importance of Being Causally Prescriptive

Abstract

This article proposes a framework for transition from traditional data science where the focus is on extracting value from available data to goal-driven analytical decision making where the business objective is defined first. We discuss the link between predictive analytics and prescriptive analytics in the context of formulating the problem, and assert that all prescriptive analytics problem formulations assume a causal link between decisions and outcomes. We emphasize the role of predictive analytics and causal inference in specifying the causal link between decisions and outcomes accurately, and ultimately in aligning the analysis with the business objectives. We offer practical examples that integrate various required analytics tasks and describe scenarios where causal inference is required versus not required.

Keywords

• prescriptive analytics
• predictive analytics
• causal inference
• data science
• decision-making