ConeOpt is a Python package that implements an optimization based conterfactual explanations for any machine learning estimator.

1. Create a Python env adn then activate it.

conda create -p ./env39 python=3.9 --yes
conda activate ./env39

2. To install reuqired packages, use:

pip install -r requirements.txt

3. To test ConeOpt, run notebook and open the file: POC2.ipynb.

jupyter notebook

4. To test Alibi, follow the instruction here to install the package. Then open alibi/examples/cfproto_mnist.ipynb to test the Counterfactual Analysis. Detailed explanation of the implementaiton can be found here.

Other related documentations can be found here.