timesead.data.transforms.artificial_anomalies

Classes

InjectArtificialAnomaliesTransform	This Transform injects anomalies into the dataset.
InjectIndependentArtificialAnomaliesTransform	Transform that injects anomalies, that only depend on the anomaly interval.
InjectWindowsArtificialAnomaliesTransform	Transform that inject windows from somewhere else in the dataset as anomalies.

Module Contents

class timesead.data.transforms.artificial_anomalies.InjectArtificialAnomaliesTransform(parent: timesead.data.transforms.transform_base.Transform, n: int, min_length: int = 1, max_length: int = 1)

Bases: timesead.data.transforms.transform_base.Transform

This Transform injects anomalies into the dataset.

It expects the get_datapoint method of its parent to return a tuple of tuples of length 1.

Parameters:

• parent (timesead.data.transforms.transform_base.Transform) – Another Transform which is used as the data source for this Transform.

• n (int) – Number of anomalies to insert.

• min_length (int) – Minimum length of anomalies.

• max_length (int) – Maximum length of anomalies.

class timesead.data.transforms.artificial_anomalies.InjectIndependentArtificialAnomaliesTransform(parent: timesead.data.transforms.transform_base.Transform, anomaly_fn: Callable, n: int, min_length: int = 1, max_length: int = 1)

Bases: InjectArtificialAnomaliesTransform

Transform that injects anomalies, that only depend on the anomaly interval.

Parameters:

• parent (timesead.data.transforms.transform_base.Transform) – Another Transform which is used as the data source for this Transform.

• anomaly_fn (Callable) – Callable that adds an anomaly to an interval and returns a torch.Tensor of the same size as its input.

• n (int) – Number of anomalies to insert.

• min_length (int) – Minimum length of anomalies.

• max_length (int) – Maximum length of anomalies.

anomaly

class timesead.data.transforms.artificial_anomalies.InjectWindowsArtificialAnomaliesTransform(parent: timesead.data.transforms.transform_base.Transform, mask_fn: Callable, n: int, min_length: int = 1, max_length: int = 1)

Bases: InjectArtificialAnomaliesTransform

Transform that inject windows from somewhere else in the dataset as anomalies.

Parameters:

• parent (timesead.data.transforms.transform_base.Transform) – Another Transform which is used as the data source for this Transform.

• mask_fn (Callable) – Callable that computes a mask to the features of an interval.

• n (int) – Number of anomalies to insert.

• min_length (int) – Minimum length of anomalies.

• max_length (int) – Maximum length of anomalies.

mask