A Guide for Practical Use of ADMG Causal Data Augmentation

TL;DR

This paper evaluates the ADMG causal data augmentation method for tabular data, highlighting its strengths and limitations in small-data regimes and providing insights for effective application.

Contribution

It offers an experimental analysis of ADMG augmentation, clarifying when prior causal knowledge improves data generation and model robustness.

Findings

ADMG is model-agnostic and independent of data mechanism.

Requires a minimal number of observations, which can be challenging in small-data settings.

Propagates outliers, degrading model performance.

Abstract

Data augmentation is essential when applying Machine Learning in small-data regimes. It generates new samples following the observed data distribution while increasing their diversity and variability to help researchers and practitioners improve their models' robustness and, thus, deploy them in the real world. Nevertheless, its usage in tabular data still needs to be improved, as prior knowledge about the underlying data mechanism is seldom considered, limiting the fidelity and diversity of the generated data. Causal data augmentation strategies have been pointed out as a solution to handle these challenges by relying on conditional independence encoded in a causal graph. In this context, this paper experimentally analyzed the ADMG causal augmentation method considering different settings to support researchers and practitioners in understanding under which conditions prior knowledge helps generate new data points and, consequently, enhances the robustness of their models. The results highlighted that the studied method (a) is independent of the underlying model mechanism, (b) requires a minimal number of observations that may be challenging in a small-data regime to improve an ML model's accuracy, (c) propagates outliers to the augmented set degrading the performance of the model, and (d) is sensitive to its hyperparameter's value.