josemacedo / data-faker

Dataset generator for Pandas

data-faker

Generate synthetic datasets which can be used directly for research or train models, using YAML specifications. Currently, only Pandas dataframes are supported as output

Installation

The package can be easily from GitHub repository installed using Python's pip utility.

Usage

Dataset generation is fairly straightforward:

import data_faker as df

spec_file = 'examples/distributions.yaml'
output = 'output.csv'
df.generate(spec_file, output)

A command line tool is also installed during the setup, which allows to generate datasets and serialize them straight from the command line:

datafaker -o output.csv examples/distributions.yaml

or:

datafaker --output-file output.csv examples/distributions.yaml

Currently the tool supports only serialization to CSV file. However, one can easily serialize the created dataset to other formats, by generating Pandas dataframe directly using generate_pandas method, and using internal pandas methods or third-party tools.

Specification

TBD

Requirements

This tool requires several other Python libraries to function:

• NumPy
• Pandas
• faker
• ruamel.yaml