Structured Evaluation of Synthetic Tabular Data

arXiv:2403.10424v1 Announce Type: new
Abstract: Tabular data is common yet typically incomplete, small in volume, and access-restricted due to privacy concerns. Synthetic data generation offers potential solutions. Many metrics exist for evaluating the quality of synthetic tabular data; however, we lack an objective, coherent interpretation of the many metrics. To address this issue, we propose an evaluation framework with a single, mathematical objective that posits that the synthetic data should be drawn from the same distribution as the observed data. …

abstract arxiv concerns cs.lg data evaluation however interpretation issue metrics privacy quality small solutions stat.ml synthetic synthetic data tabular tabular data type