alibi_detect.od.sr

Constants

logger

logger: logging.Logger = <Logger alibi_detect.od.sr (WARNING)>

Instances of the Logger class represent a single logging channel. A "logging channel" indicates an area of an application. Exactly how an "area" is defined is up to the application developer. Since an application can have any number of areas, logging channels are identified by a unique string. Application areas can be nested (e.g. an area of "input processing" might include sub-areas "read CSV files", "read XLS files" and "read Gnumeric files"). To cater for this natural nesting, channel names are organized into a namespace hierarchy where levels are separated by periods, much like the Java or Python package namespace. So in the instance given above, channel names might be "input" for the upper level, and "input.csv", "input.xls" and "input.gnu" for the sub-levels. There is no arbitrary limit to the depth of nesting.

EPSILON

EPSILON: float = 1e-08

Convert a string or number to a floating point number, if possible.

Padding

Inherits from: str, Enum

An enumeration.

Side

Inherits from: str, Enum

An enumeration.

SpectralResidual

Inherits from: BaseDetector, ThresholdMixin, ABC

Constructor

SpectralResidual(self, threshold: float = None, window_amp: int = None, window_local: int = None, padding_amp_method: typing_extensions.Literal['constant', 'replicate', 'reflect'] = 'reflect', padding_local_method: typing_extensions.Literal['constant', 'replicate', 'reflect'] = 'reflect', padding_amp_side: typing_extensions.Literal['bilateral', 'left', 'right'] = 'bilateral', n_est_points: int = None, n_grad_points: int = 5) -> None

Name	Type	Default	Description
threshold	Optional[float]	None	Threshold used to classify outliers. Relative saliency map distance from the moving average.
window_amp	Optional[int]	None	Window for the average log amplitude.
window_local	Optional[int]	None	Window for the local average of the saliency map. Note that the averaging is performed over the previous window_local data points (i.e., is a local average of the preceding window_local points for the current index).
padding_amp_method	Literal[constant, replicate, reflect]	'reflect'	Padding method to be used prior to each convolution over log amplitude. Possible values: constant
padding_local_method	Literal[constant, replicate, reflect]	'reflect'	Padding method to be used prior to each convolution over saliency map. Possible values: constant
padding_amp_side	Literal[bilateral, left, right]	'bilateral'	Whether to pad the amplitudes on both sides or only on one side. Possible values: bilateral
n_est_points	Optional[int]	None	Number of estimated points padded to the end of the sequence.
n_grad_points	int	5	Number of points used for the gradient estimation of the additional points padded to the end of the sequence.

Methods

add_est_points

add_est_points(X: numpy.ndarray, t: numpy.ndarray) -> numpy.ndarray

Pad the time series with additional points since the method works better if the anomaly point

is towards the center of the sliding window.

Name	Type	Default	Description
X	numpy.ndarray		Uniformly sampled time series instances.
t	numpy.ndarray		Equidistant timestamps corresponding to each input instances (i.e, the array should contain numerical values in increasing order).

Returns

• Type: numpy.ndarray

compute_grads

compute_grads(X: numpy.ndarray, t: numpy.ndarray) -> numpy.ndarray

Slope of the straight line between different points of the time series

multiplied by the average time step size.

Name	Type	Default	Description
X	numpy.ndarray		Uniformly sampled time series instances.
t	numpy.ndarray		Equidistant timestamps corresponding to each input instances (i.e, the array should contain numerical values in increasing order).

Returns

• Type: numpy.ndarray

infer_threshold

infer_threshold(X: numpy.ndarray, t: Optional[numpy.ndarray] = None, threshold_perc: float = 95.0) -> None

Update threshold by a value inferred from the percentage of instances considered to be

outliers in a sample of the dataset.

Name	Type	Default	Description
X	numpy.ndarray		Uniformly sampled time series instances.
t	Optional[numpy.ndarray]	None	Equidistant timestamps corresponding to each input instances (i.e, the array should contain numerical values in increasing order). If not provided, the timestamps will be replaced by an array of integers [0, 1, ... , N - 1], where N is the size of the input time series.
threshold_perc	float	95.0	Percentage of X considered to be normal based on the outlier score.

Returns

• Type: None

pad_same

pad_same(X: numpy.ndarray, W: numpy.ndarray, method: str = 'replicate', side: str = 'bilateral') -> numpy.ndarray

Adds padding to the time series X such that after applying a valid convolution with a kernel/filter

w, the resulting time series has the same shape as the input X.

Name	Type	Default	Description
X	numpy.ndarray		Time series to be padded
W	numpy.ndarray		Convolution kernel/filter.
method	str	'replicate'	Padding method to be used. Possible values:
side	str	'bilateral'	Whether to pad the time series bilateral or only on one side. Possible values:

Returns

• Type: numpy.ndarray

predict

predict(X: numpy.ndarray, t: Optional[numpy.ndarray] = None, return_instance_score: bool = True) -> Dict[Dict[str, str], Dict[numpy.ndarray, numpy.ndarray]]

Compute outlier scores and transform into outlier predictions.

Name	Type	Default	Description
X	numpy.ndarray		Uniformly sampled time series instances.
t	Optional[numpy.ndarray]	None	Equidistant timestamps corresponding to each input instances (i.e, the array should contain numerical values in increasing order). If not provided, the timestamps will be replaced by an array of integers [0, 1, ... , N - 1], where N is the size of the input time series.
return_instance_score	bool	True	Whether to return instance level outlier scores.

Returns

• Type: Dict[Dict[str, str], Dict[numpy.ndarray, numpy.ndarray]]

saliency_map

saliency_map(X: numpy.ndarray) -> numpy.ndarray

Compute saliency map.

Name	Type	Default	Description
X	numpy.ndarray		Uniformly sampled time series instances.

Returns

• Type: numpy.ndarray

score

score(X: numpy.ndarray, t: Optional[numpy.ndarray] = None) -> numpy.ndarray

Compute outlier scores.

Name	Type	Default	Description
X	numpy.ndarray		Uniformly sampled time series instances.
t	Optional[numpy.ndarray]	None	Equidistant timestamps corresponding to each input instances (i.e, the array should contain numerical values in increasing order). If not provided, the timestamps will be replaced by an array of integers [0, 1, ... , N - 1], where N is the size of the input time series.

Returns

• Type: numpy.ndarray