Metadata-Version: 2.4
Name: model_agnostic_toolkit
Version: 2.0.2
Summary: A package for model agnostic interpretable methods. Supports regression and classification.
Home-page: https://git-ce.rwth-aachen.de/wzl-iqs3/quality-insights/publications/model_agnostic_toolkit
Author: Tobias Schulze, Chrismarie Enslin, Daniel Buschmann, Marcos Padrón Hinrichs, Felix Sohnius, Robert H. Schmitt
Author-email: tobias.schulze@wzl-iqs.rwth-aachen.de
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: OS Independent
Description-Content-Type: text/markdown
License-File: LICENSE.md
Requires-Dist: bunch
Requires-Dist: jupyter
Requires-Dist: networkx
Requires-Dist: numpy==1.23.5
Requires-Dist: pandas
Requires-Dist: plotly==5.24
Requires-Dist: tables
Requires-Dist: scikit-learn<1.3,>=1.0
Requires-Dist: shap
Requires-Dist: lime==0.2.0.1
Requires-Dist: dice-ml==0.11
Requires-Dist: anchor-exp==0.0.2.0
Requires-Dist: cdt
Requires-Dist: pgmpy
Requires-Dist: pyale
Requires-Dist: torch==2.2
Requires-Dist: xgboost>=1.2
Requires-Dist: IPython
Requires-Dist: PDPbox==0.3.0
Requires-Dist: joblib
Requires-Dist: psutil
Requires-Dist: alibi
Requires-Dist: charset-normalizer==3.1.0
Requires-Dist: seaborn
Requires-Dist: ray
Provides-Extra: docs
Requires-Dist: sphinx>=7.0; extra == "docs"
Requires-Dist: myst-parser>=3.0; extra == "docs"
Requires-Dist: sphinx-rtd-theme>=2.0; extra == "docs"
Requires-Dist: sphinx-autodoc-typehints>=2.0; extra == "docs"
Requires-Dist: sphinxcontrib-mermaid>=0.9; extra == "docs"
Dynamic: author
Dynamic: author-email
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: license-file
Dynamic: provides-extra
Dynamic: requires-dist
Dynamic: summary


The Model Agnostic Toolkit is a package for explaining the predictions of machine learning
models on tabular datasets - both how the model behaves overall and why it arrived at a
specific prediction for a single instance. It includes a multitude of tools for three main
applications:

- **Feature importances** - how does a single feature affect the target, globally across the
  dataset?
- **Feature pair interactions** - what effects toward the target exist between features?
- **Local (instance-level) explanations** - why did the model arrive at its prediction for one
  specific instance?

Supported tools include PFI, PDP, VIP, ALE, the H-Statistic and a global sensitivity analysis
(DSA) for the importance and interaction use cases, as well as SHAP, LIME, DiCE, Anchors and ICE
for local explanations of individual instances. Results are available both as raw values and as
interactive Plotly visualizations.

For more details, please refer to the project documentation.
