raise-synthetic-data-generator 0.1.5

RAISE Synthetic data generator

RAISE Synthetic Data Generator

A Python package to generate shareable versions of tabular datasets ready to upload to the RAISE platform. The package can also be used beyond the scope of the RAISE project as a lightweight synthetic data generator.

Features

The package currently provides a single main function: generate_synthetic_data.

Main capabilities

• Input flexibility Accepts either:

  • A CSV file path, or
  • A pandas.DataFrame.

• Automatic or manual model selection

  • "auto-select" (default) automatically chooses the best synthetic data generation model based on the input data.
  • You can also specify a model explicitly (for the moment one of CTGAN, TVAE or Copulas).

• Synthetic data generation

  • Generates a synthetic dataset with the same properties as the input data.
  • Number of synthetic samples to generate can be specified via n_samples (defaults to the size of the input data).

• Results storage

  • Saves the generated synthetic dataset as synthetic_data.csv.
  • Stores model information (info.txt) inside the chosen output folder.
  • Creates a run-specific folder under the desired output path.

• Evaluation report (optional)

  • If evaluation_report=True (default), runs a quality assessment comparing original vs synthetic data.
  • Produces an evaluation report (evaluation_report.pdf) with figures and summary statistics.

• Logging and error handling

  • Provides informative log messages for each step (dataset loading, model selection, data generation, report creation).
  • Exceptions are logged with full traceback and re-raised for debugging.

Installation

You can install raise-synthetic-data-generator directly from PyPI using pip:

pip install raise-synthetic-data-generator

Usage

from raise_synthetic_data_generator import generate_synthetic_data
import pandas as pd

# Example input dataframe
df = pd.DataFrame(
    {"age": [23, 35, 44, 29, 31], "country": ["ES", "FR", "DE", "IT", "ES"]}
)

# Generate synthetic data (in memory + saved to disk)
generate_synthetic_data(
    dataset=df,  # if desired the CSV filename can also be given
    selected_model="auto-select",  # or explicitly: "CTGAN", "TVAE" or "Copulas
    n_samples=10,  # number of synthetic samples to generate
    evaluation_report=True,  # if true (evaluation PDF report is generated)
    output_dir="results",  # base output directory (if none, results path will be created)
    run_name="demo-run",  # optional run name (this will be the subfolder where generated objects will be stored, if none a subfolder will be created)
)

This will save in specified output folder:

• The generated synthetic (synthetic_data.csv)
• A text file with the applied model information (info.txt)
• (If selected) A folder with resulted evaluation figures (evaluation_figures).
• (If selected) A PDF report with synthetic data quality evaluation results (evaluation_report.pdf).

Usage Examples

Code examples demonstrating how to use the raise-synthetic-data-generator package are provided in the examples folder of the repository. You can explore these examples to understand how to utilize the functionality of the package.

To get started, check the examples folder for various scripts and notebooks, such as:

• generate_synthetic_data.ipynb: A Jupyter Notebook with step-by-step instructions for generating synthetic data of your dataset.

License

This project is licensed under the European Union Public License (EUPL) version 1.2. See the LICENSE file for more details.

Contributing

We welcome contributions! If you'd like to contribute, please fork the repository, make changes, and submit a pull request. Contributions are subject to the terms of the EUPL license.

Contact

For any inquiries, feel free to reach out via the following email: [e-mail]. More about the project: https://raise-science.eu