Iterate through the sliding windows of each array in `arrays`. This method is not re-entrant, i.e., calling :meth:`get_iterator`

Get the `reconstruction probability` of specified KPI observations. The larger `reconstruction probability`, the less likely a point

Replace masked elements of `x` with reconstructed outputs. This method can be used to do missing data imputation on `x`, with the recons

Get the reconstruction probability for `x` and `y`. The larger `reconstruction probability`, the less likely a point is anom

Complete `timestamp` such that the time interval is homogeneous. Zeros will be inserted into each array in `arrays`, at missing points.

Train the :class:`Donut` model with given data. Args: values (np.ndarray): 1-D `float32` array, the standardized

Standardize a Args: values (np.ndarray): 1-D `float32` array, the KPI observations. mean (float): If not :obj:`None`, will us

Generate augmented data. Args: values (np.ndarray): 1-D float32 array of shape `(data_length,)`, the sta

Iterate through all the mini-batch slices. Args: length (int): Total length of data in an epoch. batch_size (int): Size of e

Get the training loss for `x` and `y`. Args: x (tf.Tensor): 2-D `float32` :class:`tf.Tensor`, the windows of

Iteratively reconstruct `x` with `mask` for `iter_count` times. This method will call :func:`masked_reconstruct` for `iter_count` times,

Get the :class:`Donut` model instance. Returns: Donut: The :class:`Donut` model instance.

Derived classes should override this to actually implement the data augmentation algorithm.

Get the mean of the training data.

Get the standard deviation of training data.

Get the ratio of missing points to inject.

Get the :class:`Donut` model instance. Returns: Donut: The :class:`Donut` model instance.

Get the VAE object of this :class:`Donut` model. Returns: VAE: The VAE object of this model.

Get the number of `x` dimensions.

Get the number of `z` dimensions.