nshap 0.1.0

Python package to compute n-Shapley Values.

Welcome to the nshap Package!

This is a python package to compute $n$-Shapley Values.

$n$-Shapley Values are a natural extension of Shapley Values and Shapley Interaction Values and were introduced in the paper From Shapley Values to Generalized Additive Models and back.

The package works with arbitrary user-defined value functions. It also provides a model-agnostic implementation of the interventional SHAP value function.

The computed $n$-Shapley Values are an estimate that can be inaccurate.

⚠️ Disclaimer

This package does not provide an efficient way to compute Shapley Values. For this you should refer to the shap package. In practice, the current implementation works for arbitrary functions of up to ~10 variables. This package should be used for research purposes only.

Setup

To install the package run

pip install nshap

A Simple Example

Let's assume that we have trained a Gradient Boosted Tree on the Folktables Income data set.

gbtree = xgboost.XGBClassifier()
gbtree.fit(X_train, Y_train)
print(f'Accuracy: {accuracy_score(Y_test, gbtree.predict(X_test)):0.3f}')

Accuracy: 0.829

In order to compute $n$-Shapley Values, we need to define a value function. The function nshap.vfunc.interventional_shap approximates the interventional SHAP value function.

import nshap

vfunc = nshap.vfunc.interventional_shap(gbtree.predict_proba, X_train, target=0, num_samples=1000)

The function takes 4 arguments

• The function that we want to explain
• The training data or another sample from the data distribution
• The target class (required here since 'predict_proba' has 2 outputs).
• The number of samples that should be used to estimate the conditional expectation (Default: 1000)

Equipped with a value function, we can compute $n$-Shapley Values.

n_shapley_values = nshap.n_shapley_values(X_test[0, :], vfunc, n=10)

The function returns an object of type nShapleyValues. It is a python dict with some added functionallity.

To get the interaction effect between features 2 and 3, simply call

n_shapley_values[(2,3)]

0.0074

To generate the plots in the paper, call

n_shapley_values.plot(feature_names = feature_names)

and to compute 2-Shapley Values and generate a plot, use

n_shapley_values.k_shapley_values(2).plot(feature_names = feature_names)

We can also compare these results with the Shapley Values returned by the shap package.

For this, we approximate the Shapley Values with Kernel SHAP

import shap

explainer = shap.KernelExplainer(gbtree.predict_proba, shap.kmeans(X_train, 25))
shap.force_plot(explainer.expected_value[0], shap_values[0])

and then generate the same plot for the Shapley Values that we just computed with the nshap package.

shap.force_plot(vfunc(X_test[0,:], []), n_shapley_values.shapley_values())

There are slight differences which is not surprising since we used two very different methods to compute the Shapley Values.

Overview of the package

Computing $n$-Shapley Values

The most important function in the package is n_shapley_values(X, v_func, n=-1) which computes $n$-Shapley Values. It takes 3 arguments

• X: A data set or a single data point for which to compute the $n$-Shapley Values (numpy.ndarray)

• v_func: A value function, the basic primitive in the computation of all Shapley Values (see below)

• The 'n' of the $n$-Shapley Values. Defaults to the number of features (complete functional decomposition or Shapley-GAM).

The function returns a list of nShapleyValues for each data point, or a single object of type nShapleyValues if there is only a single data point.

The nShapleyValues class

The nShapleyValues class is a python dict with some added functionallity. It supports the following operations.

• The individual attributions can be indexed with tuples of integers. For example, indexing with (0,) returns the main effect of the first feature.

• plot() generates the plots described in the paper.

• k_shapley_values(k) computes the $k$-Shapley Values using the recursive relationship among $n$-Shapley Values of different order (requires $k\leq n$).

• shapley_values() returns the associated original Shapley Values as a list. Useful for compatiblity with the shap package.

• save(fname) serializes the object to json. Can be loaded from there with nshap.load(fname). This can be useful since computing $n$-Shapley Values takes time, so you might want to compute them in parallel in the cloud, then aggregate the results for analysis.

Definig Value Functions

A value function has to follow the interface v_func(x, S) where x is a single data point (a numpy.ndarray) and S is a python list with the indices the the coordinates that belong to the coaltion.

In the introductory example with the Gradient Boosted Tree,

vfunc(x, [])

returns the expected predicted probability that an observation belongs to class 0, and

vfunc(x, [0,1,2,3,4,5,6,7,8,9])

returns the predicted probability that the observation x belongs to class 0 (note that the problem is 10-dimensional).

Implementation Details

The function nshap.n_shapley_values computes $n$-Shapley Values simply via their definition. Independent of the order n of the $n$-Shapley Values, this requires to call the value function v_func once for all $2^d$ subsets of coordinates. Thus, the current implementation provides no essential speedup for the computation of $n$-Shapley Values of lower order.

The function nshap.vfunc.interventional_shap approximates the interventional SHAP value function by intervening on the coordinates of randomly sampled points from the data distributions.

Notes on Estimation

The computed $n$-Shapley Values are an estimate which can be inaccurate.

The estimation error depends on the precision of the value function. With the provided implementation of the interventional SHAP value function, the precision depends on the number of samples used to estimate the expectation.

A simple way to test whether your result is precisely estimated to increase the number of samples (the num_samples parameter of nshap.vfunc.interventional_shap) and see if the result changes.

For more details, check out the discussion in Section 8 of our paper.

Replicating the Results in our Paper

The folder notebooks\replicate-paper contains Jupyter Notebooks that allow to replicated the results in our paper.

• The notebooks figures.ipynb and checkerboard-figures.ipynb generate all the figures in the paper.
• The notebook estimation.ipynb provides the estimation example with the kNN classifier on the Folktables Travel data set that we discuss in Appendix Section B.
• The notebook hyperparameters.ipynb cross-validates the parameter $k$ of the kNN classifier.
• The notebooks compute.ipynb, compute-vfunc.ipynb, checkerboard-compute.ipynb and checkerboard-compute-million.ipynb compute the different $n$-Shapley Values. You do not have to run these notebooks, the pre-computed results can be downloaded here.

Citing nshap

If you use this software in your research, we encourage you to cite our paper.

@article{bordtlux2022,
  title={From Shapley Values to Generalized Additive Models and back},
  author={Bordt, Sebastian and von Luxburg, Ulrike},
  url = {https://arxiv.org/abs/2209.04012},
  publisher = {arXiv},
  year = {2022},
}