Dan Simpson • 9/2/2022

Priors part 4: Specifying priors that appropriately penalise complexity

This article is part of a series on prior distributions in Bayesian statistics. It critiques common prior-setting methods and introduces the concept of penalized complexity priors. The author explores the philosophical nature of parameters, compares model parameterizations (like Negative-Binomial vs. Poisson-Gamma), and argues for priors that naturally penalize unnecessary model complexity to avoid overfitting.

0 comments

#Machine Learning #Statistical Modeling #Bayesian Statistics