Method improves synthetic data quality by optimizing variable order
A goodness of fit approach to select the optimal visit sequence for nonparametric data synthesis
Machine Learning
Summary
Synthetic data helps protect privacy when sharing research information. When making synthetic data, the order in which variables are created matters a lot. The authors introduce a new way to pick this order by checking how well the generated data fits the real data. Their method works better and faster than older approaches and can be used with any kind of data and model.
What this means in practice
- •For data privacy teams: Generate higher quality synthetic datasets for sensitive research while ensuring privacy compliance by choosing optimal variable generation order.
- •For statistical software developers: Incorporate an efficient variable ordering method into synthetic data tools to improve speed and output quality across diverse data types.
Authors
R{é}my Chapelle, Bruno Falissard, Mohammed Sedki
Abstract
Synthetic data are a promising way to maintain privacy while disseminating research data, thereby enabling compliance with the open science principles. Popular methods for generating synthetic data include sequential synthesis, where each variable is generated conditionally on the preceding ones in a given order. This order, also called the visit sequence of the synthesizer, is known to affect the quality of synthetic data. This raises the question of the criteria that should guide its choice. In this work, we present a novel method to perform such selection based on a goodness of fit criterion, and show that it outperforms alternative structure-based methods, both in terms of synthesis quality and time complexity. The proposed method also has the advantage of being applicable to any type of data and any class of conditional distribution models.