Library to generate toy data for machine learning experiments in the context of all-relevant feature selection.
All Relevant Feature Selection Generator Library (ARFS-Gen)
This repository contains a python library to generate synthetic (toy) data for use in research papers when evaluating all relevant feature selection e.g. used in fri.
It allows creating datasets with a specified number of strongly and weakly relevant features as well as random noise features.
In the newest revision it also includes methods which generate data with privileged information.
It works by utilizing existing methods from
The library is available on PyPi.
pip install arfs_gen
or clone this repository and use:
pip install .
In the following we generate a simple regression data set with a mix of strongly and weakly relevant features:
# Import relevant method from arfs_gen import genRegressionData # Import model from sklearn.svm import LinearSVR # Specify parameters n = 100 # Features strRel = 2 strWeak = 2 # Overall number of features (Rest will be filled by random features) d = 10 # Generate the data X, y = genRegressionData( n_samples=n, n_features=d, n_redundant=strWeak, n_strel=strRel, n_repeated=0, noise=0, ) # Fit a model linsvr = LinearSVR() linsvr.fit(X, y)
For dependency management we use the newly released poetry tool.
If you have
poetry installed, use
$ poetry install
inside the project folder to create a new
venv and to install all dependencies.
To enter the newly created
$ poetry env
to open a new shell inside.
Or alternatively run commands inside the
poetry run ....
Test it by running
poetry run pytest.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.