# FPDE

## Docs

- [Introduction](https://fpde-80-mintlify-48090872.mintlify.site/index.md): Use FPDE to explain classification results with prototype-contrast feature attribution
- [Installation](https://fpde-80-mintlify-48090872.mintlify.site/installation.md): Install FPDE from PyPI or from a local repository checkout
- [Quickstart](https://fpde-80-mintlify-48090872.mintlify.site/quickstart.md): Train a scikit-learn classifier and explain one prediction with FPDE
- [Explain one sample](https://fpde-80-mintlify-48090872.mintlify.site/explain-one.md): Fit FPDEEngine and inspect a single target-versus-rival explanation
- [Explain batches](https://fpde-80-mintlify-48090872.mintlify.site/explain-batches.md): Generate attribution matrices and per-sample metadata with FPDEEngine
- [Select lambda_hyb](https://fpde-80-mintlify-48090872.mintlify.site/select-lambda.md): Choose a fixed Hyb-FPDE mixture weight or build a Bayesian posterior over lambda_hyb candidates using deletion and insertion curves on held-out data.
- [Validate explanations](https://fpde-80-mintlify-48090872.mintlify.site/validate-explanations.md): Use perturbation curves and sanity checks to inspect FPDE attribution quality
- [Method overview](https://fpde-80-mintlify-48090872.mintlify.site/method-overview.md): Understand prototypes, target-rival contrasts, and FPDE variants
- [Interpreting results](https://fpde-80-mintlify-48090872.mintlify.site/interpreting-results.md): Read FPDE signs, evidence values, and result metadata correctly
- [Reproducibility checklist](https://fpde-80-mintlify-48090872.mintlify.site/reproducibility.md): Record the environment, data, model, and FPDE settings behind an experiment
- [API reference](https://fpde-80-mintlify-48090872.mintlify.site/api-reference.md): Stable public FPDE 0.1.0 classes, functions, parameters, and result objects
- [Troubleshooting](https://fpde-80-mintlify-48090872.mintlify.site/troubleshooting.md): Fix common FPDE setup, model, shape, and parameter errors
- [Week of September 7, 2026](https://fpde-80-mintlify-48090872.mintlify.site/changelog/2026-09-07.md): A new runnable notebook walks through the Bayesian-FPDE workflow from posterior selection to posterior-mean explanations.
- [Week of June 8, 2026](https://fpde-80-mintlify-48090872.mintlify.site/changelog/2026-06-08.md): Bayesian-FPDE builds a posterior over lambda_hyb candidates, and a new optional plotting module visualizes FPDE attributions out of the box.
- [Week of June 1, 2026](https://fpde-80-mintlify-48090872.mintlify.site/changelog/2026-06-01.md): FPDE v0.1.0 ships on PyPI with prototype-contrast attribution, validation helpers, and MIT or Apache-2.0 dual licensing.

## Optional

- [GitHub](https://github.com/fpde-xai/fpde)
- [PyPI](https://pypi.org/project/fpde/)