Skip to content

Normalization

Overview

METIS normalizes all raw metric values to [0, 1] where 1 = best quality. This enables:

  • Comparison across metrics with different scales
  • Aggregation into composite scores
  • Cross-dataset comparability

Normalization Types

Each metric has a normalization type that determines how the raw value is transformed:

Type Raw interpretation Formula
Distance Lower = better 1 - clip((raw - lower) / (upper - lower))
Similarity Higher = better clip((raw - lower) / (upper - lower))
Bounded [0,1] Already in [0,1] clip(raw) or calibrated

Metric Normalization Map

All 48 metrics are pre-classified:

Distance metrics (lower raw = better synthetic quality)

  • KS, Wasserstein, Anderson-Darling, Hellinger, KDE-ISE
  • Delta Mean, Delta Median, Delta IQR, Delta MAD, Cohen's d
  • TVD, JS, KL, PSI, Entropy Delta, Gini Delta
  • Pearson delta, Spearman delta, MI delta, dCor delta
  • Eta², Point-Biserial, Kruskal ε²
  • Cramér's V delta, Theil's U delta, χ² statistic delta
  • Correlation Matrix, MMD, Energy Distance
  • DCR (inverted), NNAA, MIA, Record Linkage, Inference Attack

Similarity metrics (higher raw = better)

  • Outliers Coverage
  • TSTR, TRTS, TTS, TTRS, ML Efficiency
  • k-Anonymity, l-Diversity, t-Closeness
  • Differential Privacy, Delta Exceedance

Stochastic Dominance Aggregation

After normalization, scores are aggregated using First-Order Stochastic Dominance (FSD):

  1. Per-column scores are computed for each metric
  2. FSD determines if one distribution of scores dominates another
  3. Family scores (fidelity, utility, privacy) are computed
  4. A composite score combines all three dimensions

This is more robust than simple averaging because it respects the ordinal structure of scores.