How to use Synthetic Data Generation tool
Your quick start guide
Welcome to Synthetic Data Generation!
To begin, users must specify their data source, select the data format (Tabular or Timeseries), and define the desired objective (Standard Generation or Handle Imbalance).
Generation From Scratch: Create entirely new datasets by defining custom schemas and statistical parameters.
-
Tabular: Generate structured data for databases and spreadsheets.
-
Time-series: (To be added in next version)
Generation From Existing Data: Synthesize data by capturing the underlying distributions and complex correlations of a reference dataset.
-
Tabular Data: Generate synthetic replicas of structured data using advanced statistical modeling and deep learning architectures (e.g., GANs, VAEs).
-
Timeseries Data: Produce synthetic sequential data that preserves temporal dependencies and seasonal patterns utilizing specialized deep learning frameworks.
-
Imbalance Handling: Apply targeted synthesis to augment minority classes, effectively mitigating bias in skewed datasets to improve model performance.
After selecting an asset, the user must configure the required metadata and model parameters to generate the desired output. Finally, the SDG tool offers a comprehensive evaluation suite to validate the synthetic output against the original reference dataset. The tool assesses data quality by analyzing statistical correlations and employing diverse metrics to evaluate both categorical and numerical data distributions.
Steps
Step 1: Get Familiar
First thing first we need you to get a bit familiar with the tool. So press the link and jump in to start Getting Familiar.
Step 2: Configure Project
Configure your first project with us, so to start getting more familiar with the tool. Configure Project.
Step 3: Evaluation
The Evaluation provides a diagnostic environment designed to validate the quality of your synthetic data, by comparing the synthetic output against the original source. So jump in to get a better picture of it. Evaluation.