POPS Regression#

Date: Mar 23, 2026 Version: 0.4.1

Useful links: Source Repository | Issues & Ideas

Bayesian regression for low-noise data with misspecification uncertainty estimation using the POPS (Pointwise Optimal Parameter Sets) algorithm.

popsregression provides POPSRegression, a scikit-learn compatible estimator that extends BayesianRidge to estimate weight uncertainties accounting for model misspecification. This is particularly useful for surrogate models fit to near-deterministic data, where standard Bayesian regression significantly underestimates predictive uncertainty.

Getting started

Installation and quick introduction to POPS Regression.

User guide

Background on the POPS algorithm and how to use it effectively.

API reference

Detailed API documentation for POPSRegression.

Examples

Gallery of examples demonstrating POPS Regression usage.