Contents Menu Expand Light mode Dark mode Auto light/dark, in light mode Auto light/dark, in dark mode Skip to content
aggmetrics 0.2.0
aggmetrics 0.2.0
  • Installation
  • A five-minute tour
  • Pitfall gallery
    • The floor trap: a classifier wrong on every example can score 0.998
    • Micro, macro and imbalance
    • Multilabel: the best model depends on the operator
    • Ordinal grades and costs: an error of one grade is not an error of four
    • Model selection: the scorer decides, and so does how you average the folds
    • Long-tailed images: the same model under different class priors
    • Choosing a threshold: why you should not tune on recall
    • Soft labels: how much do humans disagree, and does it change the verdict?
    • Segmentation: pixel measures, tiles and empty images
    • Your own measure: what the package infers, and where it can be wrong
    • Evaluating PyTorch and Keras models
    • Choosing which measures to report
  • User guide
    • Concepts
    • Measures and their properties
    • Aggregation: operators, conventions and folds
    • Diagnostics
    • Soft labels and annotator proportions
    • Ordinal and cost-sensitive evaluation
    • Comparing models
    • Your own measures
    • Choosing which measures to report
    • What if the classes were more or less frequent?
    • Positive-unlabeled data
    • scikit-learn integration
  • API reference
    • Indicator matrices
    • Measures
    • Aggregation
    • Diagnostics
    • Property inference
    • Choosing a suite of measures
    • Prior shift
    • Several targets
    • Taxonomy cost matrices
    • Positive-unlabeled learning
    • T-norms
    • scikit-learn integration
  • For AI coding assistants
  • Theory and tests
  • Contributing
  • Changelog
Back to top
View this page
Edit this page

User guideΒΆ

  • Concepts
  • Measures and their properties
  • Aggregation: operators, conventions and folds
  • Diagnostics
  • Soft labels and annotator proportions
  • Ordinal and cost-sensitive evaluation
  • Comparing models
  • Your own measures
  • Choosing which measures to report
  • What if the classes were more or less frequent?
  • Positive-unlabeled data
  • scikit-learn integration
Next
Concepts
Previous
Choosing which measures to report
Copyright © 2026, Ronaldo C. Prati
Made with Sphinx and @pradyunsg's Furo