Principle:Sdv dev SDV Multi Table Quality Evaluation
| Knowledge Sources | |
|---|---|
| Domains | Data_Quality, Evaluation, Relational_Data |
| Last Updated | 2026-02-14 00:00 GMT |
Overview
A statistical evaluation framework that measures how faithfully synthetic multi-table relational data reproduces the distributions and relationships of real data.
Description
Multi-table quality evaluation extends single-table evaluation to relational datasets. It evaluates each table individually for column shapes and pair trends, and additionally assesses inter-table relationship quality. The framework produces a QualityReport with per-table and overall scores.
Usage
Use after generating multi-table synthetic data with HMASynthesizer to assess quality.
Theoretical Basis
The evaluation applies single-table metrics per table and aggregates across all tables. The same Column Shapes and Column Pair Trends metrics are used, applied independently to each table in the relational dataset.