AI synthesize

AI synthesize allows you to synthesize realistic data using machine learning models trained on your original dataset. This method maintains statistical fidelity while ensuring privacy and unlinkability to the source records.

When to use

  • To create synthetic datasets for machine learning or analytics

  • When high statistical accuracy and maximum privacy are required

  • To expand datasets while preserving original distributions

When not to use

  • When working with multiple related tables

  • When data consistency across systems is required

  • When you need to be able to revert to original records

  • If entirely new, unseen text values must be generated

  • If the data needs to follow specific rules with 100% certainty​

Supported data types

The Syntho platform supports a wide variety of data types. Under the hood, Syntho uses an encoding scheme where each data type is mapped to one of the following encoding types.

Data type
Description

Numerical counts (e.g. number of visits)

Continuous values (e.g. weight, temperature)

Predefined values (e.g. blood type, country)

Timestamps and dates (e.g. created at)

Interactive guide: How to apply AI synthesize

Follow the interactive guide below to apply AI synthesize.

To protect privacy, Syntho can automatically replace infrequent values in categorical columns:

  • Threshold: minimum frequency before a value is considered rare (default = 10)

  • Replacement: value used to replace rare categories (default = *)

Generator-level

  • Max rows used for training: limit data for faster performance

  • Take random sample: randomly sample rows for training

Column-level

  • Clipping thresholds: restrict extreme values in numeric/date columns

  • Locale: set language model context for text/PII

Last updated

Was this helpful?