Implementation:Sdv dev SDV BaseConstraint Is Valid

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

Overview

Concrete tool for validating synthetic data against constraint definitions, provided by the SDV library.

Description

The BaseConstraint.is_valid method checks whether data rows satisfy the constraint's rules. It accepts either a DataFrame (single-table) or a dictionary of DataFrames (multi-table) and returns a boolean Series indicating row validity.

Usage

Call is_valid on a fitted constraint with synthetic data to check satisfaction.

Code Reference

Source Location

• Repository: SDV
• File: sdv/cag/base.py
• Lines: L305-333

Signature

def is_valid(self, data, metadata=None):
    """Check whether table rows are valid for this constraint.

    Args:
        data (pd.DataFrame or dict[pd.DataFrame]): Data to validate.
        metadata (Metadata or None): Optional metadata (required if not fitted).

    Returns:
        pd.Series or dict[str, pd.Series]: Boolean values per row.
    """

Import

from sdv.cag import Inequality
# is_valid is called as: constraint.is_valid(synthetic_data)

I/O Contract

Inputs

Name	Type	Required	Description
data	pd.DataFrame or dict	Yes	Data to validate
metadata	Metadata or None	No	Required if constraint not yet fitted

Outputs

Name	Type	Description
return value	pd.Series or dict[str, pd.Series]	Boolean Series indicating valid rows

Usage Examples

from sdv.cag import Inequality

constraint = Inequality(
    low_column_name='start_date',
    high_column_name='end_date',
)

# After fitting the constraint
constraint.fit(data, metadata)

# Validate synthetic data
validity = constraint.is_valid(synthetic_data)
print(f"Valid rows: {validity.sum()} / {len(validity)}")

Related Pages

Implements Principle

• Principle:Sdv_dev_SDV_Constraint_Validation

Requires Environment

• Environment:Sdv_dev_SDV_Python_Runtime