User Guide

• 1. Foundations
  • 1.1. What Is Quantification?
    • 1.1.1. When Is Quantification the Right Tool?
  • 1.2. Why Not Just Classify and Count?
    • 1.2.1. The CC Bias
  • 1.3. Dataset Shift
    • 1.3.1. Prior Probability Shift (Label Shift)
    • 1.3.2. Covariate Shift
    • 1.3.3. Concept Shift (Concept Drift)
  • 1.4. The Aggregative Quantification Framework
    • 1.4.1. The Role of Cross-Validation in fit
    • 1.4.2. The estimator_fitted parameter
  • 1.5. Method Families at a Glance
  • 1.6. Choosing the Right Evaluation Protocol
    • 1.6.1. Choosing the Right Evaluation Metric
  • 1.7. Multiclass Quantification
  • 1.8. Further Reading
• 2. Aggregative Quantification
  • 2.1. Using Aggregative Quantification Methods
  • 2.2. General Concept
    • 2.2.1. Examples
  • 2.3. Counting-Based Quantifiers
    • 2.3.1. CC — Classify and Count
    • 2.3.2. PCC — Probabilistic Classify and Count
    • 2.3.3. GACC — Generalised Adjusted Classify and Count
    • 2.3.4. GPACC — Generalised Probabilistic Adjusted Classify and Count
    • 2.3.5. Choosing Among CC, PCC, GACC, and GPACC
  • 2.4. Counters For Quantification
    • 2.4.1. Classify and Count
    • 2.4.2. Probabilistic Classify and Count
  • 2.5. Adjusted Counting
    • 2.5.1. The Adjustment Formula
    • 2.5.2. ACC — Adjusted Classify and Count (hard predictions)
    • 2.5.3. ThresholdAdjustment — Base Class for ROC-Threshold Methods
    • 2.5.4. TAC — Threshold Adjusted Count (fixed threshold)
    • 2.5.5. TX — Threshold X (symmetric ROC point)
    • 2.5.6. TMAX — Maximum TPR−FPR Separation
    • 2.5.7. T50 — TPR ≈ 0.5 Threshold
    • 2.5.8. MS — Median Sweep
    • 2.5.9. MS2 — Median Sweep with Constraint
    • 2.5.10. Comparing Threshold-Adjustment Methods
    • 2.5.11. Threshold Adjustment
  • 2.6. Likelihood Methods
    • 2.6.1. Prior Probability Shift — The Core Assumption
    • 2.6.2. MLPE — Maximum Likelihood Prevalence Estimation (trivial baseline)
    • 2.6.3. EMQ — Expectation-Maximization Quantifier (SLD)
    • 2.6.4. CDE — CDE-Iterate (threshold-adjustment via cost ratios)
    • 2.6.5. Method Comparison
    • 2.6.6. Maximum Likelihood Prevalence Estimation (MLPE)
    • 2.6.7. Expectation Maximization for Quantification (EMQ)
  • 2.7. Distribution Matching
    • 2.7.1. Histogram Methods
    • 2.7.2. Density Methods — KDEy
    • 2.7.3. Kernel Methods — MMD
    • 2.7.4. Score Methods — SORD
    • 2.7.5. Choosing a Distribution Matching Method
    • 2.7.6. Histogram
    • 2.7.7. Density
    • 2.7.8. Kernel
    • 2.7.9. Scores
  • 2.8. Nearest Neighbours
    • 2.8.1. PWK — Probabilistic Weighted k-Nearest Neighbours
    • 2.8.2. Nearest Neighbors
    • 2.8.3. PWK: Pair-wise Weighted K-Nearest Neighbors
• 3. Non Aggregative Quantification
  • 3.1. Mixture Models for Non-Aggregative Quantification
    • 3.1.1. HDx: Hellinger Distance x-Similarity
• 4. Meta Quantification
  • 4.1. Ensembles and Adaptation
    • 4.1.1. EnsembleQ — Ensemble of Quantifiers
    • 4.1.2. QuaDapt — Adaptive Score Simulation
    • 4.1.3. AggregativeBootstrap — Confidence Intervals via Bootstrap
    • 4.1.4. Choosing a Meta-Quantifier
    • 4.1.5. Ensemble for Quantification
  • 4.2. Bootstrap in Quantification
  • 4.3. QuaDapt: Drift-Resilient Score Adaptation
• 5. Model Selection and Evaluation
  • 5.1. Evaluation Protocols
    • 5.1.1. APP — Artificial Prevalence Protocol
    • 5.1.2. NPP — Natural Prevalence Protocol
    • 5.1.3. UPP — Uniform Prevalence Protocol
    • 5.1.4. Choosing a Protocol
    • 5.1.5. Protocols for Quantification
    • 5.1.6. Artificial-Prevalence Protocol (APP)
    • 5.1.7. Natural-Prevalence Protocol (NPP)
    • 5.1.8. Uniform Prevalence Protocol (UPP)
    • 5.1.9. Personalized Prevalence Protocol (PPP)
    • 5.1.10. References
  • 5.2. Hyperparameter Tuning
    • 5.2.1. References
  • 5.3. Evaluation Metrics
    • 5.3.1. Absolute Error Metrics
    • 5.3.2. Relative Error Metrics
    • 5.3.3. Divergence Metrics
    • 5.3.4. Ordinal Metrics
    • 5.3.5. Distribution Distance Metrics
    • 5.3.6. Choosing a Metric
  • 5.4. Single Label Quantification (SLQ) Metrics
    • 5.4.1. AE (Absolute Error)
    • 5.4.2. SE (Squared Error)
    • 5.4.3. MAE (Mean Absolute Error)
    • 5.4.4. MSE (Mean Squared Error)
    • 5.4.5. KLD (Kullback-Leibler Divergence)
    • 5.4.6. RAE (Relative Absolute Error)
    • 5.4.7. NAE (Normalized Absolute Error)
    • 5.4.8. NRAE (Normalized Relative Absolute Error)
    • 5.4.9. NKLD (Normalized Kullback-Leibler Divergence)
  • 5.5. Regression-Based Quantification (RQ) Metrics
    • 5.5.1. VSE (Variance Shift Error)
    • 5.5.2. CvM_L1 (Cramér-von Mises L1 Distance)
  • 5.6. Ordinal Quantification (OQ) Metrics
    • 5.6.1. NMD (Normalized Match Distance)
    • 5.6.2. RNOD (Relative Normalized Overall Deviation)
• 6. Confidence Intervals
  • 6.1. General Concept
  • 6.2. Percentile-Based Confidence Intervals
  • 6.3. Confidence Ellipse in Simplex
  • 6.4. Confidence Ellipse in CLR Space
    • 6.4.1. References
• 7. Building a Quantifier
  • 7.1. General Quantifiers
  • 7.2. Aggregative Quantifiers
  • 7.3. Aggregation Types
  • 7.4. Binary Quantifiers
  • 7.5. Validation Utilities
  • 7.6. Parameter Constraints
• 8. Mlquantify methods
• 9. Core Components
  • 9.1. Composable Quantifiers
    • 9.1.1. Key classes
    • 9.1.2. Linear compose example
    • 9.1.3. Likelihood compose example
  • 9.2. Loss Functions
    • 9.2.1. Available losses
    • 9.2.2. Example
  • 9.3. Representations
    • 9.3.1. Core representations
    • 9.3.2. Example
  • 9.4. Solvers
    • 9.4.1. Available solvers
    • 9.4.2. Example
  • 9.5. Calibration
    • 9.5.1. Available classes
    • 9.5.2. Example skeleton
  • 9.6. Neural Quantifiers
    • 9.6.1. QuaNet — Quantification Network