opendataval.dataval.knnshap package#

Submodules#

opendataval.dataval.knnshap.knnshap module#

class opendataval.dataval.knnshap.knnshap.KNNShapley(*args, **kwargs)#

Bases: DataEvaluator, ModelLessMixin

Data valuation using KNNShapley implementation.

KNN Shapley is a model-less mixin. This means we cannot specify an underlying prediction model for the DataEvaluator. However, we can specify a pretrained embedding model.

References#

Parameters#

k_neighborsint, optional

Number of neighbors to group the data points, by default 10

batch_sizeint, optional

Batch size of tensors to load at a time during training, by default 32

embedding_modelModel, optional

Pre-trained embedding model used by DataEvaluator, by default None

random_stateRandomState, optional

Random initial state, by default None

evaluate_data_values() → ndarray#

Return data values for each training data point.

Compute data values using KNN Shapley data valuation

Returns#

np.ndarray

Predicted data values/selection for training input data point

match(y: Tensor) → Tensor#

\(1.\) for all matching rows and \(0.\) otherwise.

train_data_values(*args, **kwargs)#

Trains model to predict data values.

Computes KNN shapley data values, as implemented by the following. Ignores all positional and key word arguments.

References#

Module contents#