Sliceline is a Python library for fast slice finding for Machine Learning model debugging.
It is an implementation of SliceLine: Fast, Linear-Algebra-based Slice Finding for ML Model Debugging, from Svetlana Sagadeeva and Matthias Boehm of Graz University of Technology.
Given an input dataset X and a model error vector errors,
SliceLine finds the top slices in X that identify where a ML model
performs significantly worse.
You can use sliceline as follows:
from sliceline.slicefinder import Slicefinder
slice_finder = Slicefinder()
slice_finder.fit(X, errors)
print(slice_finder.top_slices_)
X_trans = slice_finder.transform(X)We invite you to check the demo notebooks for a more thorough tutorial:
- Implementing Sliceline on Titanic dataset
- Implementing Sliceline on California housing dataset
Sliceline is intended to work with Python 3.7 or above. Installation
can be done with pip:
pip install slicelineThere are wheels available for Linux, MacOS, and Windows, which means that you most probably won’t have to build Sliceline from source.
You can install the latest development version from GitHub as so:
pip install git+https://github.com/DataDome/sliceline --upgradeOr, through SSH:
pip install git+ssh://[email protected]/datadome/sliceline.git --upgradeFeel free to contribute in any way you like, we’re always open to new ideas and approaches.
- Open a discussion if you have any question or enquiry whatsoever. It’s more useful to ask your question in public rather than sending us a private email. It’s also encouraged to open a discussion before contributing, so that everyone is aligned and unnecessary work is avoided.
- Feel welcome to open an issue if you think you’ve spotted a bug or a performance issue.
Please check out the contribution guidelines if you want to bring modifications to the code base.
Sliceline is free and open-source software licensed under the 3-clause BSD license.