14.4 What is a model?

• A data science model is:
  • A representation of the world built from random variables
  • FOIS: “agnostic” models place minimal restrictions on joint distribution
  • Parametric models (i.e. MLE) are models based on a family of distributions.
  • \(f_{Y|X}(y|\mathbf{x}) \sim g(y, \mathbf{x}; \mathbf{\theta})\)