Baselines

class dtaianomaly.anomaly_detection.baselines.AlwaysNormal

Baseline anomaly detector, which predicts that all observations are normal. This detector should only be used for sanity-check, and not to effectively detect anomalies in time series data.

decision_function(X: ndarray) → ndarray

Predicts ‘always normal’ anomaly scores, i.e., always returns a ‘0.0’.

Parameters:

X (array-like of shape (n_samples, n_attributes)) – Input time series.

Returns:

normal_scores – All normal anomaly scores.

Return type:

array-like of shape (n_samples)

Raises:

ValueError – If X is not a valid array.

fit(X: ndarray, y: ndarray | None = None) → AlwaysNormal

Simply return this detector, because no fitting is required.

Parameters:

• X (array-like of shape (n_samples, n_attributes)) – Input time series.

• y (ignored) – Not used, present for API consistency by convention.

Returns:

self – Returns the instance itself

Return type:

AlwaysNormal

class dtaianomaly.anomaly_detection.baselines.AlwaysAnomalous

Baseline anomaly detector, which predicts that all observations are anomalous. This detector should only be used for sanity-check, and not to effectively detect anomalies in time series data.

decision_function(X: ndarray) → ndarray

Predicts ‘always anomalous’ anomaly scores, i.e., always returns a ‘1.0’.

Parameters:

X (array-like of shape (n_samples, n_attributes)) – Input time series.

Returns:

anomalous_scores – All anomalous anomaly scores.

Return type:

array-like of shape (n_samples)

Raises:

ValueError – If X is not a valid array.

fit(X: ndarray, y: ndarray | None = None) → AlwaysAnomalous

Simply return this detector, because no fitting is required.

Parameters:

• X (array-like of shape (n_samples, n_attributes)) – Input time series.

• y (ignored) – Not used, present for API consistency by convention.

Returns:

self – Returns the instance itself

Return type:

AlwaysAnomalous

class dtaianomaly.anomaly_detection.baselines.RandomDetector(seed: int | None = None)

Baseline anomaly detector, which assigns random anomaly scores. This detector should only be used for sanity-check, and not to effectively detect anomalies in time series data.

Parameters:

seed (int, default=None) – The seed to use for generating anomaly scores. If None, no seed will be used.

decision_function(X: ndarray) → ndarray

Predicts random anomaly scores. Uses numpy random-number generator, without adjusting the internal seed of numpy.

Parameters:

X (array-like of shape (n_samples, n_attributes)) – Input time series.

Returns:

random_scores – Random anomaly scores.

Return type:

array-like of shape (n_samples)

Raises:

ValueError – If X is not a valid array.

fit(X: ndarray, y: ndarray | None = None) → RandomDetector

Simply return this detector, because no fitting is required.

Parameters:

• X (array-like of shape (n_samples, n_attributes)) – Input time series.

• y (ignored) – Not used, present for API consistency by convention.

Returns:

self – Returns the instance itself

Return type:

RandomDetector