# scikit-learn-contrib/MAPIE

A scikit-learn-compatible library for estimating prediction intervals and controlling risks, based on conformal predictions.

Repository: https://github.com/scikit-learn-contrib/MAPIE
Canonical: https://ross.abutalabs.com/products/mapie
Homepage: https://mapie.readthedocs.io/en/stable/
Language: Jupyter Notebook
License: BSD-3-Clause
License Family: permissive
Topics: regression, confidence-intervals, data-science, python, sklearn, classification, conformal-prediction, risk-control
Last push: 2026-08-14T13:49:29+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 97, release rhythm 84, longevity 100
- inputs: {"age_days": 1982, "days_push": 19, "days_rel": 28, "gap_med": 57, "n_releases_24m": 12}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1584, forks 151 (observed 2026-08-28T04:05:07.425518+00:00)

## What it is
MAPIE is a scikit-learn-compatible Python library for quantifying uncertainty in machine learning models via conformal prediction. It computes prediction intervals and prediction sets for regression, classification, and time series, and provides risk-control mechanisms.

## Use cases
- estimate prediction intervals for regression models
- compute prediction sets for classification with coverage guarantees
- quantify uncertainty in time series forecasts
- control false positive rates of ML models
- apply risk control to LLM-as-Judge evaluations
- verify exchangeability before applying conformal prediction

## When to choose
- you use scikit-learn and need calibrated uncertainty estimates
- you need statistical coverage guarantees for predictions
- you want model-agnostic conformal prediction methods

## When to avoid
- you need Bayesian uncertainty methods rather than conformal ones
- your model or data violates exchangeability assumptions
- you work outside the Python/scikit-learn ecosystem

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, testing
- domain: machine-learning, data-science
- platform: python
- tags: conformal-prediction, prediction-intervals, uncertainty-quantification, risk-control, scikit-learn-compatible

## Member repositories
- scikit-learn-contrib/MAPIE (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:07.425518+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:55:58.810065+00:00, confidence not recorded.
  - readme: https://github.com/scikit-learn-contrib/MAPIE (fetched 2026-08-28T04:05:07.425518+00:00, sha 15bb03e1edc7)
  - registry_pypi: https://pypi.org/pypi/mapie/json (fetched 2026-08-29T11:27:05.595234+00:00, sha 6df105754055)
- Data as of 2026-08-30T08:39:29.467469+00:00.