netcal.presentation.ReliabilityRegression

class netcal.presentation.ReliabilityRegression(quantiles: int | Iterable[float] | ndarray = 11)

Reliability diagram in the scope of regression calibration for probabilistic regression models. This diagram visualizes the quantile coverage frequency for several quantile levels and plots these observed coverage scores above the desired quantile levels. In this way, it is possible to compare the predicted and the observed quantile levels with each other.

This method is able to visualize the quantile coverage in terms of multiple univariate distributions if the input is given as multiple independent Gaussians. This method is also able to visualize the multivariate quantile coverage for a joint multivariate Gaussian if the input is given with covariance matrices.

Parameters:

quantiles (int or iterable, default: 11) – Quantile levels that are used for the visualization of the regression reliability diagram. If int, use NumPy’s linspace method to get the quantile levels. If iterable, use the specified quantiles for visualization.

Methods

__init__([quantiles])

Constructor.

plot(X, y, *[, kind, filename, tikz, ...])

Reliability diagram for regression calibration to visualize the predicted quantile levels vs.