ADMAntwerp/NICE

Nearest Instance Counterfactual Explanations (NICE)

NICE is an algorithm to generate Counterfactual Explanations for heterogeneous tabular data. Our approach exploits information from a nearest instance to speed up the search process and guarantee that an explanation will be found.

Installation

Install NICE through Pypi

pip install NICEx

or github

pip install git git+https://github.com/ADMantwerp/nice.git 

Usage

NICE requires acces to the prediction score and trainingdata to generate counterfactual explanations.

from nice import NICE

# Initialize NICE by specifing the optimization strategy and providing the training data and predictive model.
NICE_explainer = NICE(
    X_train=X_train,
    predict_fn=predict_fn,
    y_train=y_train,
    cat_feat=cat_feat,
    num_feat=num_feat
)

# explain an instance
NICE_explainer.explain(x)

Examples

NICE on Adult

References

NICE: An Algorithm for Nearest Instance Counterfactual Explanations