Macro-F1 Measure from Group Counts Method

A Macro-F1 Measure from Group Counts Method is a macro-averaged performance measure computation method that averages per-group F1 scores computed from aligned true positive counts, false positive counts, and false negative counts across multiple classification groups.

• AKA: Group-Averaged F1 Computation Method, Unweighted Average F1 Method, Macro F1 from Counts Method, Class-Averaged F1 Score Method.
• Context:
  • It can typically aggregate F1 Measure from Counts Method results across classification groups.
  • It can typically assume Independent Groups Assumption in Variance Estimation Method for variance calculations.
  • It can typically weight all group F1 scores equally regardless of group support.
  • It can often serve as input to Macro-F1 P-Value Calculation Methods.
  • It can often enable Macro-F1 Difference P-Value Method for model comparisons.
  • It can often handle class imbalance through equal weighting schemes.
  • It can range from being a Simple Macro-F1 Measure from Group Counts Method to being a Weighted Macro-F1 Measure from Group Counts Method, depending on its group weighting scheme.
  • It can range from being a Binary Macro-F1 Measure from Group Counts Method to being a Multi-Class Macro-F1 Measure from Group Counts Method, depending on its number of groups.
  • It can range from being a Static Macro-F1 Measure from Group Counts Method to being a Dynamic Macro-F1 Measure from Group Counts Method, depending on its group definition stability.
  • It can range from being a Complete Macro-F1 Measure from Group Counts Method to being a Partial Macro-F1 Measure from Group Counts Method, depending on its group coverage.
  • It can range from being a Deterministic Macro-F1 Measure from Group Counts Method to being a Stochastic Macro-F1 Measure from Group Counts Method, depending on its sampling approach.
  • ...
• Example(s):
  • Three-Class Macro-F1 Calculations, such as:
    • Class A: F1=0.8, Class B: F1=0.7, Class C: F1=0.9, Macro-F1=0.8.
    • Imbalanced classes with equal macro weighting.
  • Multi-Label Macro-F1 Calculations, such as:
    • Averaging F1 across multiple binary labels.
    • Document classification with 20 categories.
  • Cross-Validation Macro-F1s, such as:
    • Averaging macro-F1 across 5 folds.
    • Stratified k-fold with group preservation.
  • ...
• Counter-Example(s):
  • Micro-F1 Measure Method, which aggregates counts before computing F1.
  • Weighted F1-Measure Method, which uses class support weights.
  • Single-Class F1 Method, which doesn't aggregate across groups.
• See: Macro-F1 Measure, F1 Measure from Counts Method, Performance Measure Computation Method, Unweighted Arithmetic Mean, Independent Groups Assumption in Variance Estimation Method, Macro-F1 P-Value Calculation Method, Macro-F1 Difference P-Value Method, Delta-Method F1 Standard Error Estimation Method, Micro-F1 Measure from Group Counts Method, Weighted F1 Measure from Group Counts Method, Per-Class F1 Score, Multi-Class Classification Task, Class Imbalance Problem.

---