/datacleaning-benchmark

Primary LanguagePythonMIT LicenseMIT

What is this?

This is a set of python decorators that inject a wide range of noise into your machine learning pipelines.

See test/testcleaningbenchmark.py for example code.

Installation

Install

    pip install cleaningbenchmark

    # or

    python setup.py install

Test it

    python test/testcleaningbenchmark.py

Documentation

Documentation coming soon.

Usage

    # create noise model 
    g = EscapeCorruptionModel([0], 0.5)  

    # augment pipeline_operation to inject escape corruption noise to its first argument
    @inject(g)
    def pipeline_operation(data, args):
      return data

List of Noise Models

TBA