Welcome to PyCalib’s documentation!

PyCalib is a library that offers multiple tools to assess probabilistic classifiers in terms of calibration, and provides helpful functions to calibrate muticlass probabilistic models that follow the Scikit-learn BaseEstimator standard.

Contents:

• Installation
• Quickstart
• Calibration methods
  • BinningCalibration
  • CalibratedModel
  • IsotonicCalibration
  • LogisticCalibration
  • SigmoidCalibration
• Metrics
  • ECE()
  • MCE()
  • accuracy()
  • binary_ECE()
  • binary_MCE()
  • brier_score()
  • classwise_ECE()
  • classwise_MCE()
  • conf_ECE()
  • conf_MCE()
  • cross_entropy()
• Visualisations
  • plot_reliability_diagram()
• Statistics tools
  • TestResult
  • compute_friedmanchisquare()
• Examples

Indices and tables

• Index

• Module Index