Towards a framework on tabular synthetic data generation: a minimalist approach: theory, use cases, and limitations

TL;DR

This paper introduces a minimalist framework for generating synthetic tabular data using simple, interpretable models that outperform complex autoencoders in robustness testing and practical applications.

Contribution

The paper presents a novel minimalist approach combining SparsePCA and XGBoost for synthetic data generation, emphasizing simplicity, interpretability, and robustness over autoencoder-based methods.

Findings

The method is effective in high-dimensional credit scoring data.

It provides a practical alternative for robustness testing.

The approach requires no extra tuning and maintains interpretability.

Abstract

We propose and study a minimalist approach towards synthetic tabular data generation. The model consists of a minimalistic unsupervised SparsePCA encoder (with contingent clustering step or log transformation to handle nonlinearity) and XGboost decoder which is SOTA for structured data regression and classification tasks. We study and contrast the methodologies with (variational) autoencoders in several toy low dimensional scenarios to derive necessary intuitions. The framework is applied to high dimensional simulated credit scoring data which parallels real-life financial applications. We applied the method to robustness testing to demonstrate practical use cases. The case study result suggests that the method provides an alternative to raw and quantile perturbation for model robustness testing. We show that the method is simplistic, guarantees interpretability all the way through, does not require extra tuning and provide unique benefits.