{
  "name": "Multiclass Averaging Metric Response Across Imbalance Ratios",
  "description": "Empirical comparison of scikit-learn classification evaluation metrics and averaging behaviors across varying levels of cohort class imbalance, compiled from scikit-learn model evaluation benchmarks and scientific literature.",
  "page": "https://casrai.org/guides/sklearn-f1-score",
  "publisher": "CASRAI",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "dateModified": "2026-11-15T17:55:04Z",
  "temporalCoverage": "2026",
  "spatialCoverage": "Global",
  "columns": [
    {
      "name": "Imbalance Profile",
      "unit": "",
      "description": "Classification cohort distribution profile"
    },
    {
      "name": "Class Ratio",
      "unit": "",
      "description": "Ratio of majority to minority instances"
    },
    {
      "name": "Minority Class Prevalence",
      "unit": "%",
      "description": "Percentage of total cohort belonging to the smallest target class"
    },
    {
      "name": "Accuracy Robustness",
      "unit": "",
      "description": "Reliability of raw classification accuracy on this distribution"
    },
    {
      "name": "Recommended F1 Averaging",
      "unit": "",
      "description": "Recommended scikit-learn f1_score average parameter"
    },
    {
      "name": "Macro-F1 Sensitivity",
      "unit": "",
      "description": "Sensitivity of macro-averaged F1 to minority class misclassifications"
    }
  ],
  "rows": [
    {
      "Imbalance Profile": "Balanced Benchmark",
      "Class Ratio": "1:1:1",
      "Minority Class Prevalence": 33.3,
      "Accuracy Robustness": "High reliability",
      "Recommended F1 Averaging": "macro or micro",
      "Macro-F1 Sensitivity": "Moderate"
    },
    {
      "Imbalance Profile": "Moderate Clinical Imbalance",
      "Class Ratio": "5:3:2",
      "Minority Class Prevalence": 20,
      "Accuracy Robustness": "Moderate distortion",
      "Recommended F1 Averaging": "macro",
      "Macro-F1 Sensitivity": "High"
    },
    {
      "Imbalance Profile": "Severe Epidemiological Skew",
      "Class Ratio": "10:1:1",
      "Minority Class Prevalence": 8.3,
      "Accuracy Robustness": "Severe inflation",
      "Recommended F1 Averaging": "macro",
      "Macro-F1 Sensitivity": "Critical"
    },
    {
      "Imbalance Profile": "Extreme Rare Pathology",
      "Class Ratio": "50:5:1",
      "Minority Class Prevalence": 1.8,
      "Accuracy Robustness": "Extremely misleading",
      "Recommended F1 Averaging": "macro or None",
      "Macro-F1 Sensitivity": "Maximum"
    }
  ],
  "sources": [
    {
      "title": "Scikit-learn User Guide: Model Evaluation",
      "url": "https://scikit-learn.org/stable/modules/model_evaluation.html#precision-recall-f-measure-metrics",
      "accessed": "2026-11-15"
    },
    {
      "title": "Pedregosa et al. (2011) Scikit-learn: Machine Learning in Python",
      "url": "https://jmlr.csail.mit.edu/papers/v12/pedregosa11a.html",
      "accessed": "2026-11-15"
    }
  ]
}