
Data Science: Causation and Leverage

Marketplace lessons from Uber, Airbnb, Bumble, and more | Ramesh Johari (Stanford professor)
Topic 3 of 6
Published Nov 9, 2023 · Clip 26:07–40:22 (14:16)
Ramesh highlights that data scientists should prioritize causation over correlation, focusing on decisions' impacts rather than mere predictions. In marketplaces, leverage comes from evaluating search rankings, match quality, and feedback via experiments, not just predictive models. He distinguishes prediction (patterns in past data) from decision-making (differential effects), using examples like hiring predictions and LTV promotions.