GPyConform

GPyConform extends the GPyTorch library by implementing Conformal Prediction (CP) for Gaussian Process Regression (GPR), providing distribution-free, finite-sample valid Prediction Intervals (PIs) under the sole assumption of data exchangeability.

GPyConform supports both the Transductive (Full) CP and Inductive (Split) CP versions of the framework through a unified interface. In both cases it implements a GPR-specific normalized nonconformity measure [1] that leverages the GP predictive variance to construct adaptive symmetric or asymmetric conformal prediction intervals.

Key Features

• Provably Valid Prediction Intervals: Distribution-free, finite-sample coverage guarantees under minimal assumptions (data exchangeability).
• Two CP Framework Versions:
  • Transductive (Full) CP for Exact GPs: ExactGPCP
  • Inductive (Split) CP for any GPyTorch regression model: GPRICPWrapper, plus a model-agnostic InductiveConformalRegressor
• Symmetric and Asymmetric PIs in both frameworks.
• Normalized Nonconformity that leverages the GP predictive variance for tighter, adaptive intervals.
• Unified PI Container + Metrics: PredictionIntervals supports retrieving intervals at multiple confidence levels and evaluating empirical coverage error and interval widths.
• Torch-native + GPU-friendly: Works directly with torch.Tensors and can leverage GPU acceleration.

Note

• Transductive CP (ExactGPCP) targets ExactGP models with GaussianLikelihood and relies on an internal patch to GPyTorch’s DefaultPredictionStrategy (applied automatically by default). You can control patching via the GPYCONFORM_AUTOPATCH environment variable, or call gpyconform.apply_patches() manually.
• Inductive CP does not modify the model internals and can be used with any GPyTorch regression model (including approximate/deep GPs and different likelihoods). InductiveConformalRegressor can also be used with non-GPyTorch regressors that provide predictive means/variances.

Documentation

For detailed documentation and usage examples, see GPyConform Documentation.

Installation

From PyPI

pip install gpyconform

From conda-forge

conda install conda-forge::gpyconform

Citing GPyConform

If you use GPyConform for a scientific publication, you are kindly requested to cite the following paper:

Harris Papadopoulos. "GPyConform: Conformal Prediction with Gaussian Process Regression in Python". In: K. An Nguyen, Z. Luo (eds), The Importance of Being Learnable, Lecture Notes in Computer Science, vol. 16290, pp. 449–466. Springer, 2026. DOI: 10.1007/978-3-032-15120-9_20.

Bibtex entry:

@Inbook{gpyconform,
author="Papadopoulos, Harris",
editor="An Nguyen, Khuong and Luo, Zhiyuan",
title="GPyConform: Conformal Prediction with Gaussian Process Regression in Python",
bookTitle="The Importance of Being Learnable: Essays Dedicated to Alexander Gammerman",
year="2026",
publisher="Springer Nature Switzerland",
address="Cham",
pages="449--466",
isbn="978-3-032-15120-9",
doi="10.1007/978-3-032-15120-9_20",
url="https://doi.org/10.1007/978-3-032-15120-9_20"
}

For the Gaussian Process Regression Conformal Prediction approach and nonconformity measure, please also cite:

Harris Papadopoulos. "Guaranteed Coverage Prediction Intervals with Gaussian Process Regression", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 12, pp. 9072-9083, Dec. 2024. DOI: 10.1109/TPAMI.2024.3418214. (arXiv version)

Bibtex entry:

@ARTICLE{gprcp,
  author={Papadopoulos, Harris},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  title={Guaranteed Coverage Prediction Intervals with Gaussian Process Regression}, 
  year={2024},
  volume={46},
  number={12},
  pages={9072-9083},
  doi={10.1109/TPAMI.2024.3418214}
}

References

<a id="1">[1]</a> Harris Papadopoulos. "Guaranteed Coverage Prediction Intervals with Gaussian Process Regression", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 12, pp. 9072-9083, Dec. 2024. DOI: 10.1109/TPAMI.2024.3418214. (arXiv version)

<a id="2">[2]</a> Vladimir Vovk, Alexander Gammerman, and Glenn Shafer. Algorithmic Learning in a Random World, 2nd Ed. Springer, 2023. DOI: 10.1007/978-3-031-06649-8.

<a id="3">[3]</a> Harris Papadopoulos. "GPyConform: Conformal Prediction with Gaussian Process Regression in Python". In: K. An Nguyen, Z. Luo (eds), The Importance of Being Learnable, Lecture Notes in Computer Science, vol. 16290, pp. 449–466. Springer, 2026. DOI: 10.1007/978-3-032-15120-9_20.

---

Author: Harris Papadopoulos (h.papadopoulos@frederick.ac.cy) / Copyright 2024-2026 Harris Papadopoulos / License: BSD 3 clause