Full Citation

Title: Comparative Evaluation of Synthetic Data Generation Methods

Citation Type: Miscellaneous

Publication Year: 2019

ISBN:

ISSN:

DOI:

NSFID:

PMCID:

PMID:

Abstract: Unrestricted availability of datasets is important for researchers and decision makers to evaluate their strategies to solve certain problems. Equally important is the privacy of the respective data owners. Synthetically generated datasets provide a way to retain the utility of the original data keeping the privacy of the owners invulnerable. We perform a comparative study of synthetic data generation techniques using different data synthesizers like decision tree, ran- dom forest and neural network. We evaluate the effectiveness of these methods towards the amount of utility they preserve and the risk of disclosure.

Url: https://pdfs.semanticscholar.org/84ba/a6078344b5e8c7638c5f351ada4d7ea8fff6.pdf

User Submitted?: No

Authors: Dandekar, Ashish; Zen, Remmy, AM; Bressan, Stéphane

Publisher: National University of Singapore

Data Collections: IPUMS International

Topics: Methodology and Data Collection, Population Data Science

Countries: