2.6.0

Release Notes (What's New)

Version 2.6.0 (July 17, 2026)

For a list of all changes in this release, see the full changelog. Below are the changes we think users may wish to be aware of.

Features

• scores has introduced support for pandas version 3. See PR #1062.
• Added three new metrics:
  • Relative economic value: scores.continuous.relative_economic_value, scores.probability.relative_economic_value, scores.categorical.relative_economic_value and scores.plotdata.relative_economic_value. See PR #999 and PR #1088.
  • Relative economic value from rates: scores.continuous.relative_economic_value_from_rates, scores.probability.relative_economic_value_from_rates , scores.categorical.relative_economic_value_from_rates and scores.plotdata.relative_economic_value_from_rates. See PR #999.
  • Receiver (relative) operating characteristic area under curve (ROC AUC): scores.probability.roc_auc. This is significantly computationally more efficient than calculating the area under the curve using scores.probability.roc_curve_data. See PR #1036.
• Added a kwarg to scores.continuous.kge to allow switching between the original Kling-Gupta Efficiency (KGE) formulation of Gupta et al. (2009) and the modified formulation of Kling et al. (2012). The default remains the original implementation. See PR #1069.

Deprecations

• This deprecation was first introduced in Version 2.5.0. Support for include_components will be removed from threshold-weighted continuous ranked probability score (twCRPS) functions in a future version of scores. The scores development team believe using include_components=True may lead to misleading results when used with twCRPS functions. As such, the following are now deprecated:
  • support for include_components in scores.probability.tw_crps_for_ensemble,
  • support for include_components in scores.probability.tail_tw_crps_for_ensemble and
  • support for include_components in scores.probability.interval_tw_crps_for_ensemble.
    See PR #991.

Bug Fixes

• Improved NaN handling in scores.probability.roc_curve_data. See PR #1036.
• Improved dimension reduction code to ensure consistent dimension ordering in returned objects in scores.probability.roc_curve_data. See PR #1036.

Documentation

• Added "Relative Economic Value (REV)" tutorial. See PR #999.
• Updated the "Receiver Operating Characteristic (ROC)" tutorial to include information about the newly-added scores.probability.roc_auc function (which is significantly more computationally efficient). See PR #1036.
• Updated the "Kling–Gupta Efficiency (KGE)" tutorial to include information about the newly-added kwarg which allows users to switch between the original KGE formulation of Gupta et al. (2009) and the modified formulation of Kling et al. (2012). See PR #1069.
• Updated the "Contributing Guide" to include guidelines for generative tool usage. See PR #1027.
• Added a commit template to the scores repository and added instructions in the "Contribuing Guide" for setting up and using the commit template. See PR #1045 and PR #1048.
• Added an entry for "Aggregate" to the "Processing" table in docs/included.md. See PR #1038.
• Updated docstrings (e.g. added examples and improved grammar) for multiple functions in the API documentation. See PR #996 and PR #1056.
• Corrected erroneous namespaces and added :py:func: before the correct namespaces in src/scores/probability/crps_impl.py. Specifically, changed scores.probability.functions.fill_cdf to :py:func:`scores.processing.cdf.fill_cdf` and changed scores.probability.functions.cdf_envelope to :py:func:`scores.processing.cdf.cdf_envelope` . See PR #1050.
• Updated links to the new verification site https://jwgfvr.github.io/forecastverification (which will replace the prior site: https://www.cawcr.gov.au/projects/verification) in tutorials/Additive_and_multiplicative_bias.ipynb, tutorials/Binary_Contingency_Scores.ipynb and src/scores/categorical/contingency_impl.py. See PR #1029, PR #1030 and PR #1031.
• Updated link from https://www.openradar.io/ (which is no longer active) to https://nci.org.au/aura/ in docs/data.md. See PR #1081.
• Replaced link to ERA5 dataset in docs/data.md from https://www.ecmwf.int/en/forecasts/dataset/ecmwf-reanalysis-v5 to https://doi.org/10.24381/cds.adbb2d47. See PR #1081.
• Updated the "Quantile Interval Score and Interval Score" tutorial to use the updated lower-case "h" date range syntax, due to changes in pandas. See PR #1065.
• Fixed integration of tutorials with Binder. See PR #1083.

Internal Changes

• Moved tutorial files from a top-level directory, tutorials, into the docs/tutorials/ subdirectory. This change was made to support how recent Sphinx versions handle symbolic links to files located outside of the docs subdirectory. Users who run the tutorials will need to use the updated location. There is no change to how the tutorials render in the documentation. Unpinned the version of Sphinx used by scores, so that the most recent versions of Sphinx can be used. See PR #1042.
• Introduced the use of the Python doctest module for automated testing of examples in API documentation (docstrings). Revised existing docstrings as appropriate to meet doctest tool requirements. Added the doctest tool to CI/CD and developer tooling via pre-commit. See PR #1056.
• Update CI pipeline to run tests against all versions of Python even if one fails. See PR #1063.

Contributors to this Release

Oisín M. Morrison* (@Oisin-M), Daniel Karney* (@danielkarney), Thomas C. Pagano (@thomaspagano), Tennessee Leeuwenburg (@tennlee), Stephanie Chong (@Steph-Chong), Nicholas Loveday (@nicholasloveday), John Sharples (@John-Sharples), Mohammadreza Khanarmuei (@reza-armuei), Nikeeth Ramanathan (@nikeethr), Maree Carroll (@mareecarroll) and Durga Shrestha (@durgals).

* indicates that this release contains their first contribution to scores.