Describe the workflow you want to enable

import numpy as np
from sklearn.calibration import calibration_curve

y_true = np.array([0, 0, 0, 0, 1, 1, 1, 1, 1])
y_pred = np.array([0.1, 0.2, 0.3, 0.4, 0.65, 0.7, 0.8, 0.9,  1.])
prob_true, prob_pred = calibration_curve(y_true, y_pred, strategy="pav")

Describe your proposed solution

Add the strategy PAV as [1] and [2] to calibration_curve, there called CORP. This is basically applying isotonic regression as binning strategy, which we already have in scikit-learn.

[1] Dimitriadis, T., Gneiting, T., & Jordan, A.I. (2021). Stable reliability diagrams for probabilistic classifiers. Proceedings of the National Academy of Sciences of the United States of America, 118. https://doi.org/10.1073/pnas.2016191118
[2] https://cran.r-project.org/package=reliabilitydiag

Describe alternatives you've considered, if relevant

No response

Additional context

Given the recency of the paper, it has clearly not enough citations (yet). But I have the impression that this is a good strategy for a reliability diagrams with good theoretical and practical properties.

To my knowledge, this strategy is nowhere available in the Python universe as of now.