-
Causal Impact for Product Data Science
Notes on causal inference for product data science — difference-in-differences, Bayesian structural time series, and what it's actually like using Google's CausalImpact package without attribution data
-
The K-Means Paradox
Why k-means is widely used in production despite its hardest problem being cluster interpretation
-
The Abstraction Pattern
How AI is following the same pattern of abstraction that defined programming languages
-
Liouville's Theorem and the Extra Dimension
Why bounded and differentiable doesn't mean constant on the real line, but does on the complex plane
-
The Feedback Problem in Machine Learning
When your model shapes the very data it learns from