Sie können Bookmarks mittels Listen verwalten, loggen Sie sich dafür bitte in Ihr SLUB Benutzerkonto ein.
Medientyp:
E-Artikel
Titel:
Regularized Aggregation of One-Off Probability Predictions
Beteiligte:
Satopää, Ville A.
Erschienen:
Institute for Operations Research and the Management Sciences (INFORMS), 2022
Erschienen in:
Operations Research, 70 (2022) 6, Seite 3558-3580
Sprache:
Englisch
DOI:
10.1287/opre.2021.2224
ISSN:
1526-5463;
0030-364X
Entstehung:
Anmerkungen:
Beschreibung:
How much can rational people really disagree? If we can understand the limits of such disagreement, can we remove noise by labeling excess disagreement as irrational and then construct a group belief based on everyone's rational beliefs? Based on this idea, “Regularized Aggregation of One-Off Probability Predictions” by Satopää proposes a Bayesian aggregator that requires no user intervention and can be computed efficiently even for a large number of one-off probability predictions. To illustrate, the aggregator is evaluated on predictions collected during a four-year forecasting tournament sponsored by the U.S. intelligence community. The aggregator improves the squared error (a.k.a., the Brier score) of simple averaging by around 20% and other commonly used aggregators by 10%−25%. This advantage stems almost exclusively from improved calibration. An R package called braggR implements the method and is available on CRAN.