Skip to content

Conversation

@sjedhoff
Copy link

Previously, only calibration_ecdf supported custom test quantities. This PR adds that capability to additional diagnostic metrics and plotting functions.

Updated Metrics:

  • calibration_error
  • calibration_log_gamma
  • posterior_contraction
  • rott_mean_squared_error

Updated Plots:

  • calibration_histogram
  • coverage
  • recovery
  • z_score_contraction

No breaking changes; existing behavior is preserved when no custom test quantities are provided.

@paul-buerkner paul-buerkner self-requested a review October 15, 2025 12:43
@codecov
Copy link

codecov bot commented Oct 15, 2025

Codecov Report

❌ Patch coverage is 91.66667% with 5 lines in your changes missing coverage. Please review.

Files with missing lines Patch % Lines
bayesflow/diagnostics/plots/recovery.py 28.57% 5 Missing ⚠️
Files with missing lines Coverage Δ
bayesflow/diagnostics/metrics/calibration_error.py 100.00% <100.00%> (ø)
...sflow/diagnostics/metrics/calibration_log_gamma.py 100.00% <100.00%> (ø)
...sflow/diagnostics/metrics/posterior_contraction.py 100.00% <100.00%> (ø)
...low/diagnostics/metrics/root_mean_squared_error.py 60.00% <100.00%> (+7.05%) ⬆️
...yesflow/diagnostics/plots/calibration_histogram.py 92.50% <100.00%> (+1.59%) ⬆️
bayesflow/diagnostics/plots/coverage.py 97.61% <100.00%> (+0.11%) ⬆️
bayesflow/diagnostics/plots/z_score_contraction.py 96.66% <100.00%> (+1.01%) ⬆️
bayesflow/diagnostics/plots/recovery.py 82.05% <28.57%> (-11.70%) ⬇️

... and 2 files with indirect coverage changes

Copy link
Contributor

@paul-buerkner paul-buerkner left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good. Thank you!

@paul-buerkner paul-buerkner merged commit d13da56 into bayesflow-org:dev Oct 16, 2025
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants