Save Validation Results

Use the save_validation method to export Constrainer validation results as CSV files, including summary statistics, violation statistics, and detailed violation records.

Usage Example

Click the button below to run the example in Colab:

Loader:
  load_benchmark_with_schema:
    filepath: benchmark://adult-income
    schema: benchmark://adult-income_schema
Constrainer:
  external_constraints:
    method: auto
    constraints_yaml: adult-income_constraints.yaml
Reporter:
  save_validation:
    method: save_validation     # Required: Fixed to save_validation method
    # output: petsard           # Optional: Output file name prefix (default: petsard)
    # include_details: true     # Optional: Include detailed violation records (default: true)

Main Parameters

Required Parameters

Optional Parameters

include_details Parameter Details

The include_details parameter controls whether to output detailed violation record files:

1. Enable detailed records (true):

   include_details: true
   • Generates an additional *_details.csv file
   • Contains specific data rows for each violation
   • Shows up to 10 violation examples per rule
   • Includes columns: Constraint Type, Rule, Violation Index, and actual data fields

2. Disable detailed records (false):

   include_details: false
   • Only generates summary and statistics files
   • Saves storage space
   • Suitable for scenarios requiring only statistical data

Output Format

This Reporter generates up to 3 CSV files for each Constrainer validation result:

1. Summary File

Filename format:

{output}[Validation]_Constrainer[experiment_name]_summary.csv

Example filenames:

• Single source: petsard_[Validation]_Constrainer[validate_constraints]_summary.csv
• Multiple sources: petsard_[Validation]_Source[synthetic_data]_Constrainer[validate_constraints]_summary.csv

File content:

2. Violations Statistics File

Filename format:

{output}[Validation]_Constrainer[experiment_name]_violations.csv

File content:

Example:

3. Detailed Violation Records File

Filename format:

{output}[Validation]_Constrainer[experiment_name]_details.csv

Description:

• Generated only when include_details: true
• Shows up to 10 violation examples per rule
• Contains complete violation data row content

File content:

Example:

Filename Naming Rules

Using default output (petsard)

The system automatically generates complete filenames based on experiment type:

1. Single source Constrainer:

   petsard_[Validation]_Constrainer[experiment_name]_[type].csv

   Example: petsard_[Validation]_Constrainer[validate_constraints]_summary.csv

2. Multiple source Constrainer (with specified source):

   petsard_[Validation]_Source[source_name]_Constrainer[experiment_name]_[type].csv

   Example: petsard_[Validation]_Source[synthetic_data]_Constrainer[validate_constraints]_summary.csv

Using custom output

Directly uses the specified name:

Reporter:
  save_validation:
    method: save_validation
    output: my_validation_report

Output files:

• my_validation_report_summary.csv
• my_validation_report_violations.csv
• my_validation_report_details.csv (if include_details is true)

Use Cases

1. Validation Report Generation: Export data validation results as formal reports
2. Quality Monitoring: Continuously track data quality metrics
3. Problem Diagnosis: Identify data issues through detailed violation records
4. Compliance Documentation: Provide data compliance proof documents

Common Questions

Q: What’s the difference between save_validation and save_report?

A:

• save_validation: Specifically exports Constrainer validation results, generating structured CSV reports (summary, statistics, detailed records)
• save_report: Exports utility metric evaluation results, supporting multiple formats (CSV, Markdown, HTML)

Q: Why aren’t my validation results being saved?

A: Check:

1. Have you executed Constrainer’s validate method before Reporter
2. Did Constrainer successfully generate validation results
3. Is the validation result data structure correct

Q: Will too many detailed violation records affect performance?

A:

• The system automatically limits to 10 violation examples per rule
• If detailed records aren’t needed, set include_details: false to save space
• Summary and statistics files are always generated, unaffected by this setting

Q: Can I adjust the number of violation examples per rule?

A: Currently the system is fixed at 10 examples per rule. For more records, consider:

1. Using save_data to export complete validation result data
2. Or combining with Constrainer’s return_details=True option to process manually

Notes

• File Overwrite: Files with the same name will be overwritten, please backup as needed
• Data Volume Control: Detailed violation records are automatically limited to 10 examples per rule
• Encoding: All CSV files use UTF-8 encoding
• Case Insensitivity: The method parameter is case-insensitive (save_validation, SAVE_VALIDATION, or Save_Validation all work)
• Execution Order: Must be executed after Constrainer’s validate method
• Empty Value Handling: If a validation result is empty or invalid, that result will be skipped with a warning logged