shapiq.explainer.KNNExplainer

class shapiq.explainer.KNNExplainer(model, class_index=None, data=None, index='SV', max_order=1)

Bases: NNExplainerBase

Explainer for unweighted KNN models.

Implements the algorithm proposed by Jia et al.[1] to efficiently calculate Shapley values for unweighted KNN models. The algorithm itself has a linear time complexity, but requires sorting training points by distance to the test point, resulting in a time complexity of \(O(N \log N)\) for explaining a single data point.

References

[1]

Ruoxi Jia, David Dao, Boxin Wang, Frances Ann Hubis, Nezihe Merve Gurel, Bo Li, Ce Zhang, Costas J Spanos, and Dawn Song. Efficient task-specific data valuation for nearest neighbor algorithms. arXiv preprint arXiv:1908.08619, 2019. URL: https://doi.org/10.48550/arXiv.1908.08619.

Initializes the class.

Parameters:

• model (KNeighborsClassifier) – The NN model to explain. Must be an instance of sklearn.neighbors.KNeighborsClassifier or sklearn.neighbors.RadiusNeighborsClassifier. The model must not use multi-output classification, i.e. the y value provided to model.fit(X, y) must be a 1D vector.

• class_index (int | None) – The class index of the model to explain. Note that, as opposed to most Explainers, this must not be None!

• data (np.ndarray | None)

• index (ValidNNExplainerIndices)

• max_order (int)

Raises:

sklearn.exceptions.NotFittedError – The constructor was called with a model that hasn’t been fitted.

explain_function(x)

Explain a single prediction in terms of interaction values.

Parameters:

• x (ndarray[tuple[Any, ...], dtype[floating]]) – A numpy array of a data point to be explained.

• *args – Additional positional arguments passed to the explainer.

• **kwargs – Additional keyword-only arguments passed to the explainer.

Return type:

InteractionValues

Returns:

The interaction values of the prediction.

model: KNeighborsClassifier

The model to be explained, either as a Model instance or a callable function.