26 points•5 comments•1h ago•by Murfalo•fitelson.org ↗

The well-calibrated Bayesian [pdf] (1982)

Dawid defines forecast calibration as the long-run agreement between the forecast probabilities and observed frequencies, using weather forecasting as the main illustration. In a sequential, feedback-driven setting, he shows that a coherent Bayesian forecaster should expect to be well calibrated, a result with implications for coherence, and he presents a general calibration theorem extending prior work.

Discussion (5)

No comments yet

Top
Trending
New
For You
Bookmarks
Dossier