v1.0.1

Highlights

• Support for polars version 0.19.0 that introduced some breaking changes and several deprecation warnings.

What's Changed

• DOC add t-test for score difference in quantile example by @lorentzenchr in #96
• CI rename ci to deploy_docs and set workflow_dispatch trigger by @lorentzenchr in #97
• CI add 2nd monthly run of test matrix by @lorentzenchr in #100
• FIX deprecation warnings of polars 0.19 groupby and replace by @lorentzenchr in #101
• FIX edge case in decompose for constant y by @lorentzenchr in #102
• FIX isotonic_regression with median by @lorentzenchr in #103
• ENH improve y_obs and marginal in decompose by @lorentzenchr in #104
• REL increase to version 1.0.1 by @lorentzenchr in #105

Full Changelog: v1.0.0...v1.0.1

v1.0.0

Highlights

• Isotonic regression for all expectiles and quantiles enable reliability diagrams for those functionals and the score decomposition for all scoring functions that are consistent for them, e.g. for the pinball loss.
• New example on quantile regression
• The function decompose can now deal with multiple predictions at once, similar to compute_bias.

What's Changed

• DOC exclude test directories in gen_reg_pages.py by @lorentzenchr in #68
• MNT update to polars 0.17.2 by @lorentzenchr in #73
• MNT update mkdocs-material 9.1 and mkdocs-jupyter 0.24 by @lorentzenchr in #70
• ENH add generalized PAVA for isotonic regression by @lorentzenchr in #74
• ENH add IsotonicRegression class by @lorentzenchr in #78
• ENH add quantile to reliability diagram by @lorentzenchr in #79
• ENH add quantiles and expectile to scoring.decompose by @lorentzenchr in #81
• MNT update gitignore excluding some ipynb by @lorentzenchr in #82
• MNT update dependencies by @lorentzenchr in #83
• CI add monthly run of test matrix by @lorentzenchr in #84
• MNT deploy docs on release instread of on push by @lorentzenchr in #85
• MNT increase mypy to version 1.4 by @lorentzenchr in #86
• ENH add confidence_level to plot_bias by @lorentzenchr in #87
• ENH support multiple y_pred models in scoring.decompose by @lorentzenchr in #88
• DOC add quantile regression example by @lorentzenchr in #91
• DOC extend readme and index by @lorentzenchr in #93
• REL increase to version 1.0.0rc0 by @lorentzenchr in #92
• DOC update index, readme and quantile regression example by @lorentzenchr in #94
• REL increase to version 1.0.0 by @lorentzenchr in #95

Full Changelog: v0.2.0...v1.0.0

v1.0.0 release candiate 0

What's Changed

• DOC exclude test directories in gen_reg_pages.py by @lorentzenchr in #68
• MNT update to polars 0.17.2 by @lorentzenchr in #73
• MNT update mkdocs-material 9.1 and mkdocs-jupyter 0.24 by @lorentzenchr in #70
• ENH add generalized PAVA for isotonic regression by @lorentzenchr in #74
• ENH add IsotonicRegression class by @lorentzenchr in #78
• ENH add quantile to reliability diagram by @lorentzenchr in #79
• ENH add quantiles and expectile to scoring.decompose by @lorentzenchr in #81
• MNT update gitignore excluding some ipynb by @lorentzenchr in #82
• MNT update dependencies by @lorentzenchr in #83
• CI add monthly run of test matrix by @lorentzenchr in #84
• MNT deploy docs on release instread of on push by @lorentzenchr in #85
• MNT increase mypy to version 1.4 by @lorentzenchr in #86
• ENH add confidence_level to plot_bias by @lorentzenchr in #87
• ENH support multiple y_pred models in scoring.decompose by @lorentzenchr in #88
• DOC add quantile regression example by @lorentzenchr in #91
• DOC extend readme and index by @lorentzenchr in #93
• REL increase to version 1.0.0rc0 by @lorentzenchr in #92

Full Changelog: v0.2.0...v1.0.0rc0

v0.2.0

What's Changed

• DOC hyperlink in highlights of frontpage by @lorentzenchr in #46
• CI codecov setup by @lorentzenchr in #47
• DOC add codecov badge by @lorentzenchr in #48
• DOC add link to release notes by @lorentzenchr in #49
• MNT add .codecov.yml by @lorentzenchr in #50
• TST add test_compute_bias_1d_array_like by @lorentzenchr in #51
• CI upload coverage report only once by @lorentzenchr in #52
• FIX p-value when stderr is zero but bias_count>=2 by @lorentzenchr in #54
• ENH keep null values in compute_bias for low n_bins by @lorentzenchr in #53
• DOC add trunk-based development by @lorentzenchr in #55
• DOC fix typos by @mayer79 in #56
• FIX account for Null value when binning in compute_bias by @lorentzenchr in #57
• FIX fix logic for string features with n_bins in compute_bias by @lorentzenchr in #58
• ENH plot null values in plot_bias by @lorentzenchr in #59
• ENH add ElementaryScore by @lorentzenchr in #60
• FIX ElementaryScore always non-negative by @lorentzenchr in #62
• ENH add plot_murphy_diagram by @lorentzenchr in #63
• ENH plot_bias with Null values by @lorentzenchr in #65
• DOC add Murphy plot to regression example by @lorentzenchr in #66
• REL increase to version 0.2.0 by @lorentzenchr in #67

New Contributors

• @mayer79 made their first contribution in #56

Full Changelog: v0.1.1...v0.2.0

v0.1.1

Enhancements:

• Support NaN and Null in compute_bias (PR #43)

Bug Fixes:

• Always output column "bias_weights" in compute_bias (PR #44)

v0.1.0

Some highlights:

• Confidence intervals for plot_reliability_diagram via arguments n_bootstrap and confidence_level (PR #32).
• New option diagram_type = "bias" for plot_reliability_diagram, which is roughly a 45 degree rotated plot (PR #35).
• Better visualisation of uncertainty/standard errors in plot_bias and distinction between numerical and categorical features (PR #37).
• Consistently sorted output, i.e. the different (model) columns of y_pred (PR #37).
• Number of effective (output) bins is now always at most n_bins in compute_bias and plot_bias (PR #37).
• Switch to ruff (PR #34)

v0.0.3

A new module scoring containing:

• Add strictly consistent, homogeneous scoring functions
  • HomogeneousExpectileScore for mean an expectiles
  • HomogeneousQuantileScore for quantiles
  • SquaredError, PoissonDeviance, GammaDeviance and PinballLoss for convenience
• Add LogLoss
• Add score decomposition decompose 🚀
  To my knowledge, this is the first time the score decomposition into miscalibration, discrimination (or resolution) is available in Python. R users can use the wonderful reliabilitydiag package of @aijordan for quite some time now.

v0.0.2

• Added support for case weights.
• Use of the fantastic https://www.pola.rs/ library (instead of pyarrow and pandas).

v0.0.1

First public release