Model-based synthesis #2
Description
Add a model synthesizer consisting of the following components:
- An order determiner that figures out in what order the columns should be synthesized. Bayesian networks may be of use: https://www.probabilisticworld.com/bayesian-belief-networks-part-1/
- A generator that samples values of one column from a predetermined distribution.
- An inferer that learns to estimate values of one column from one or more previously synthesized columns.
- A noisifier that adds noise to the inferred values according to the error of the inferer to enhance variability and privacy.
The structure should be modular enough to enable using different models and methods for each of these components.