Implementation:Sdv dev SDV Evaluate Quality Multi Table

Knowledge Sources	SDV SDMetrics Documentation
Domains	Data_Quality, Evaluation
Last Updated	2026-02-14 00:00 GMT

Overview

Wrapper for the SDMetrics multi-table QualityReport, providing evaluation of relational synthetic data quality within SDV workflows.

Description

The multi-table evaluate_quality function wraps sdmetrics.reports.multi_table.quality_report.QualityReport. It accepts dictionaries of DataFrames and multi-table Metadata.

Usage

Call after generating multi-table synthetic data.

Code Reference

Source Location

• Repository: SDV
• File: sdv/evaluation/multi_table.py
• Lines: L10-30

Signature

def evaluate_quality(real_data, synthetic_data, metadata, verbose=True):
    """Evaluate the quality of synthetic multi-table data.

    Args:
        real_data (dict): Dictionary of real table data.
        synthetic_data (dict): Dictionary of synthetic table data.
        metadata (Metadata): Multi-table metadata.
        verbose (bool): Print report. Defaults to True.

    Returns:
        QualityReport: Multi table quality report.
    """

Import

from sdv.evaluation.multi_table import evaluate_quality

I/O Contract

Inputs

Name	Type	Required	Description
real_data	dict[str, pd.DataFrame]	Yes	Real multi-table data
synthetic_data	dict[str, pd.DataFrame]	Yes	Synthetic multi-table data
metadata	Metadata	Yes	Multi-table metadata
verbose	bool	No	Print summary (default: True)

Outputs

Name	Type	Description
return value	QualityReport	Multi-table quality report

Usage Examples

from sdv.evaluation.multi_table import evaluate_quality

quality_report = evaluate_quality(real_data, synthetic_data, metadata)
print(f"Overall quality: {quality_report.get_score()}")

Related Pages

Implements Principle

• Principle:Sdv_dev_SDV_Multi_Table_Quality_Evaluation

Requires Environment

• Environment:Sdv_dev_SDV_Python_Runtime