dataeval.detectors.drift.DriftMMD
=================================

.. py:class:: dataeval.detectors.drift.DriftMMD(x_ref, p_val = 0.05, x_ref_preprocessed = False, update_x_ref = None, preprocess_fn = None, sigma = None, configure_kernel_from_x_ref = True, n_permutations = 100, device = None)



   :term:`Maximum Mean Discrepancy (MMD) Drift Detection` algorithm using a permutation test.

   :param x_ref: Data used as reference distribution.
   :type x_ref: ArrayLike
   :param p_val: :term:`P-value` used for significance of the statistical test for each feature.
                 If the FDR correction method is used, this corresponds to the acceptable
                 q-value.
   :type p_val: float | None, default 0.05
   :param x_ref_preprocessed: Whether the given reference data ``x_ref`` has been preprocessed yet.
                              If ``True``, only the test data ``x`` will be preprocessed at prediction time.
                              If ``False``, the reference data will also be preprocessed.
   :type x_ref_preprocessed: bool, default False
   :param update_x_ref: Reference data can optionally be updated using an UpdateStrategy class. Update
                        using the last n instances seen by the detector with LastSeenUpdateStrategy
                        or via reservoir sampling with ReservoirSamplingUpdateStrategy.
   :type update_x_ref: UpdateStrategy | None, default None
   :param preprocess_fn: Function to preprocess the data before computing the data drift metrics.
                         Typically a :term:`dimensionality reduction<Dimensionality Reduction>` technique.
   :type preprocess_fn: Callable | None, default None
   :param sigma: Optionally set the internal GaussianRBF kernel bandwidth. Can also pass multiple
                 bandwidth values as an array. The kernel evaluation is then averaged over
                 those bandwidths.
   :type sigma: ArrayLike | None, default None
   :param configure_kernel_from_x_ref: Whether to already configure the kernel bandwidth from the reference data.
   :type configure_kernel_from_x_ref: bool, default True
   :param n_permutations: Number of permutations used in the permutation test.
   :type n_permutations: int, default 100
   :param device: Device type used. The default None uses the GPU and falls back on CPU.
                  Can be specified by passing either 'cuda', 'gpu' or 'cpu'.
   :type device: str | None, default None


   .. py:method:: predict(x)

      Predict whether a batch of data has drifted from the reference data and then
      updates reference data using specified strategy.

      :param x: Batch of instances.
      :type x: ArrayLike

      :returns: Output class containing the :term:`drift<Drift>` prediction, :term:`p-value<P-Value>`,
                threshold and MMD metric.
      :rtype: DriftMMDOutput



   .. py:method:: score(x)

      Compute the :term:`p-value<P-Value>` resulting from a permutation test using the maximum mean
      discrepancy as a distance measure between the reference data and the data to
      be tested.

      :param x: Batch of instances.
      :type x: ArrayLike

      :returns: p-value obtained from the permutation test, MMD^2 between the reference and test set,
                and MMD^2 threshold above which :term:`drift<Drift>` is flagged
      :rtype: tuple(float, float, float)