Demo Overview¶
The examples/demos folder contains runnable scenarios that demonstrate SynthoHive's core capabilities. Each demo is self-contained and generates its own sample data.
Setup¶
Install the package from the repository root:
pip install -e .
Each demo has a run.py script and an outputs/ folder with sample artifacts.
Available Demos¶
| Demo | Description | Spark Required? |
|---|---|---|
| 01 - Single Table CTGAN | Train CTGAN on a single table with mixed data types. | No |
| 02 - Privacy Sanitization | Detect and sanitize PII, compare raw vs sanitized outputs. | No |
| 03 - Validation Report | Generate validation metrics and HTML/JSON reports comparing real vs synthetic. | No |
| 04 - Relational Linkage CTGAN | Train relational CTGAN with linkage, synthesize users/orders with FK integrity. | No |
| 05 - Transformer Embeddings | Demonstrate DataTransformer encoding behavior and round-trip recovery. | No |
Running a Demo¶
# From the repository root
python examples/demos/01_single_table_ctgan/run.py
python examples/demos/02_privacy_sanitization/run.py
python examples/demos/03_validation_report/run.py
python examples/demos/04_relational_linkage_ctgan/run.py
python examples/demos/05_transformer_embeddings/run.py
Jupyter Notebooks¶
Interactive notebook versions of each demo are available in the notebooks/ directory. These provide step-by-step walkthroughs with inline output.