This repository contains the code for our paper:
Nonlinear Axiomatic Attribution for Cooperative Games, which can be found at https://openreview.net/forum?id=Fr6rTQZ2l3.

---

Testing and Verification

The test/ directory contains a script used to verify the correctness of our implementation.

Replicate our Results

Follow the steps below to train the models, generate the experimental results, and reproduce the figures from the paper.

1. Train Models

First, train all the required (gradient boosting) decision trees:

python createTreeModel.py

2. Run Experiments

Use the -p flag followed by the number of available CPUs to parallelize the runs:

python main.py -p <number_of_cpus>

3. Generate Figures

Once the results are ready, the figures can be plotted using:

python plot_comparison.py
python plot_best.py
python plot_heatmap.py