Inductive Venn-Abers and related regressors
2026-05-07 • Machine Learning
Machine Learning
AI summaryⓘ
The authors built on a type of predictor called Venn-Abers predictors, which were previously only used for yes/no problems and limited types of number predictions. They expanded these predictors to work for any type of number prediction without limits, using a method called conformal prediction. Their tests showed that their new approach can make better predictions than usual methods when there is a lot of training data. This helps make more reliable probabilistic predictions in complex regression tasks.
Venn-Abers predictorsprobabilistic predictorsbinary classificationregressionunbounded regressionconformal predictionpredictive efficiencypoint regressorstraining set
Authors
Ivan Petej, Vladimir Vovk
Abstract
Venn-Abers predictors are probabilistic predictors that enjoy appealing properties of validity, but their major limitation is that they are applicable only to the case of binary classification, with a recent extension to bounded regression. We generalize them to the case of unbounded regression, which requires adding an element of conformal prediction. In our simulation and empirical studies we investigate the predictive efficiency of point regressors derived from Venn-Abers regressors and argue that they somewhat improve the predictive efficiency of standard regressors for larger training sets.