R-squared for Bayesian regression models

Andrew Gelman, Ben Goodrich, Jonah Gabry, Aki Vehtari

Research output: Contribution to journalArticleScientificpeer-review

73 Citations (Scopus)
249 Downloads (Pure)

Abstract

The usual definition of R 2 (variance of the predicted values divided by the variance of the data) has a problem for Bayesian fits, as the numerator can be larger than the denominator. We propose an alternative definition similar to one that has appeared in the survival analysis literature: the variance of the predicted values divided by the variance of predicted values plus the expected variance of the errors.
Original languageEnglish
Pages (from-to)1-6
Number of pages6
JournalAMERICAN STATISTICIAN
DOIs
Publication statusPublished - 1 Jan 2019
MoE publication typeA1 Journal article-refereed

Keywords

  • Bayesian methods
  • R-squared
  • Regression

Fingerprint Dive into the research topics of 'R-squared for Bayesian regression models'. Together they form a unique fingerprint.

Cite this