A talk about data synthesis for the 2023 Strategic Data Project Convening (Chicago, IL).

This repo contains the materials for the slide deck that will be presented at SDP Convening and some examples that illustrate some of the techniques discussed. If there are questions that I am unable to address during the session, I will likely try to update the slide deck to address them. Additionally, I will likely add other synthesis examples to help illustrate other approaches/techniques if there is enough interest.

• mlSynthesisExample.Rmd
  • Contains an example of using the synthpop library to create synthetic cross-sectional data.
• ctganSynthesisExample.ipynb
  • Contains an example of using Conditional Tabular Generative Adversarial Networks (CTGAN) to create synthetic cross-sectional data.
  • Also illustrates how to generate synthetic samples for specific groups.
• cparSynthesisExample.ipynb
  • Contains an example of using Conditional Probabilistic Auto-Regressive GAN to synthesize longitudinal records.