Imagine a world where the very foundation of how we understand data is questioned. In this episode, we dive into an explosive article by Leo Breiman, who challenges the traditional "Data Modelling Culture" that statisticians have clung to for decades. Breiman argues that this obsession with data models has led to irrelevant theories and misleading conclusions, blocking statisticians from tackling more interesting, real-world problems.
Instead, he calls for a revolution—a shift to "Algorithmic Modelling Culture," where the goal is not to understand the data's mechanism, but to accurately predict outcomes using tools like decision trees and random forests. Through vivid examples, Breiman shows how these models can predict everything from ozone levels to trial delays, proving they’re more suited for today’s complex world. But here’s where it gets really intriguing: the article sparked a debate among some of the biggest names in statistics, like D.R. Cox and Brad Efron.
Are algorithmic models truly the future, or is there still value in the old ways? What do you think is the right approach to make sense of the increasingly complex data we face today?

Reference
Breiman, L. (2001). Statistical modeling: The two cultures (with comments and a rejoinder by the author). Statistical science, 16(3), 199-231.
DOI https://doi.org/10.1214/ss/1009213726

Ver todos los episodios