Synthetic Data Blueprint (SDB): A Modular Framework for Evaluating Synthetic Tabular Data

This paper introduces Synthetic Data Blueprint (SDB), a Python library designed to evaluate the fidelity of synthetic tabular data. The core problem addressed is the lack of standardized and comprehensive methods for assessing synthetic data quality. SDB offers a modular approach, incorporating feature-type detection, fidelity metrics, structure preservation scores, and data visualization. The framework's applicability is demonstrated across diverse real-world use cases, including healthcare, finance, and cybersecurity. The strength of SDB lies in its ability to provide a consistent, transparent, and reproducible benchmarking process, addressing the fragmented landscape of synthetic data evaluation. This research contributes significantly to the field by offering a practical tool for ensuring the reliability and utility of synthetic data in various AI applications.