The Synthetic Data Generation Plugin Settings

The Synthetic Data Generation Plugin Settings

Version 1.1.67.28 (Apr 15, 2026) and later

This article provides information on configuring and using the Synthetic Data Generation plugin.

Overview

The Synthetic Data Generation plugin creates synthetic datasets based on patterns identified in existing source data.

The plugin uses the Synthetic Data Vault (SDV) Python library to analyze the structure and statistical characteristics of the source dataset and generate new records with similar properties.

The generated data can be used for testing, demonstrations, development, and other scenarios where real business data should not be used.

The plugin can help:

• Create realistic test datasets without copying original records.
• Increase or reduce the number of records available for testing.
• Preserve statistical patterns and relationships between fields.
• Replace sensitive source values with newly generated values.
• Prepare datasets for testing migration, transformation, and validation logic.

The current plugin version supports one source table at a time.

Prerequisites

1. You have created a new project.

2. You have added:

• A data source plugin.
• The Synthetic Data Generation plugin.
• A data target plugin.

3. The source plugin contains the dataset that will be used as the basis for synthetic data generation.

4. The required source, plugin, and target fields have been created and mapped.

The data source can be any supported source plugin, such as Excel Import or SQL Read.

Generated data can be passed to any compatible target plugin, such as Excel Export or SQL Read & Modify.

Project structure

A typical project consists of the following plugins:

1. A source plugin that provides the original dataset.

2. The Synthetic Data Generation plugin that analyzes the source data and creates synthetic records.

3. A target plugin that stores or processes the generated dataset.

For example:

Excel Import → Synthetic Data Generation → Excel Export

The input fields of the Synthetic Data Generation plugin must be mapped from the source plugin. Its output fields must then be mapped to the selected target plugin.

Configuration

To configure the Synthetic Data Generation plugin:

1. Add the plugin to the project Design area.

2. Open the Configuration tab.

3. Specify the number of synthetic records to generate.

4. Save the configuration.

The specified value determines the number of records returned by the plugin during project execution.

Field settings

The plugin contains separate Input and Output field blocks.

Input fields

Input fields receive the source data that the plugin uses to train the synthetic data model.

Create an input field for each source column that should be included in synthetic data generation.

When creating the fields:

1. Specify the field name.

2. Select the corresponding data type.

3. Add the field.

4. Map the field from the source plugin.

The field names and data types should correspond to the source dataset.

Output fields

Output fields return the generated synthetic values.

Create an output field corresponding to each input field that should be included in the generated dataset.

When creating the fields:

1. Specify the field name.

2. Select the corresponding data type.

3. Add the field.

4. Map the field to the target plugin.

For a direct replacement of the source dataset with synthetic data, the input and output field structures should normally match.

Example: If the input structure contains Document Number, Document Type, Class, and Characteristic, create corresponding output fields for the same values.

The field configuration approach follows the standard Plugin Builder structure used for defining input and output fields. Similar Python-based plugins also require users to define fields and map them to the surrounding source and target plugins.

Business rules

Business rules are not required for the standard Synthetic Data Generation scenario.

The plugin uses its predefined Python and SDV configuration to analyze the source dataset and generate the requested number of synthetic records.

Advanced control of data generation through YAML-based business rules is not included in the current plugin scope.

Mapping

After configuring the plugin, map the fields as follows:

1. Map fields from the source plugin to the corresponding Input fields of the Synthetic Data Generation plugin.

2. Map the plugin's Output fields to the corresponding fields of the target plugin.

3. Confirm that:

• Required source fields are mapped.
• Input and output field types are compatible.
• The target plugin is configured to receive the generated data.

The plugin can be used between different combinations of supported source and target plugins.

How synthetic data is generated

During execution, the plugin:

1. Receives the source dataset through its input fields.

2. Analyzes the dataset structure and patterns.

3. Trains a synthetic data model using the source records.

4. Generates the configured number of new records.

5. Returns the generated values through its output fields.

6. Passes the generated dataset to the mapped target plugin.

Depending on the source data, the generated dataset may preserve characteristics such as:

• Data types.
• Value distributions.
• Relationships between columns.
• Common value combinations.
• Unique identifiers and key-like values.
• Statistical correlations between fields.

Sensitive values are replaced with newly generated values rather than copied directly from the source.

Running the project

After the plugin has been configured and all required fields have been mapped:

1. Save the project.

2. Analyze the project configuration.

3. Start the project.

4. Monitor the execution on the Data Upload page.

5. Review the generated dataset in the configured target.

For example, when Excel Export is used as the target, open the generated Excel file and verify that:

• The expected number of records was created.
• All mapped columns are present.
• Generated values use the expected data types.
• The output does not contain unintended copies of sensitive source values.

Limitations

The current Synthetic Data Generation plugin has the following limitations:

• Only one table can be processed at a time.
• Multi-table and relational dataset generation is not supported in the current scenario.
• Sequential dataset generation is not supported.
• Model evaluation and visual quality reports are not available through the plugin interface.
• Advanced generation settings cannot currently be controlled through business rules.
• The quality of generated data depends on the structure, consistency, and volume of the source dataset.
• A small or highly inconsistent source dataset may produce less representative synthetic values.