pintervals: an R package for model-agnostic prediction intervals

TL;DR

The pintervals R package offers a unified, model-agnostic framework for constructing and calibrating prediction intervals using various methods, facilitating comparison and application across diverse models.

Contribution

It introduces a consistent interface for multiple prediction interval methods, enabling model-agnostic construction and calibration within R.

Findings

Provides conformal, parametric, and bootstrapped prediction intervals

Supports any model with point predictions

Facilitates comparison across different prediction interval methods

Abstract

The \pkg{pintervals} package aims to provide a unified framework for constructing prediction intervals and calibrating predictions in a model-agnostic setting using set-aside calibration data. It comprises routines to construct conformal as well as parametric and bootstrapped prediction intervals from any model that outputs point predictions. Several R packages and functions already exist for constructing prediction intervals, but they often focus on specific modeling frameworks or types of predictions, or require manual customization for different models or applications. By providing a consistent interface for a variety of prediction interval construction approaches (all model-agnostic), \pkg{pintervals} allows researchers to apply and compare them across different modeling frameworks and applications.