AnomalyDetectionCatalog Class

Definition

Class used by MLContext to create instances of anomaly detection components, such as trainers and evaluators.

public sealed class AnomalyDetectionCatalog : Microsoft.ML.TrainCatalogBase
type AnomalyDetectionCatalog = class
    inherit TrainCatalogBase
Public NotInheritable Class AnomalyDetectionCatalog
Inherits TrainCatalogBase
Inheritance
AnomalyDetectionCatalog

Properties

Trainers

The list of trainers for anomaly detection.

Methods

ChangeModelThreshold<TModel>(AnomalyPredictionTransformer<TModel>, Single)

Creates a new AnomalyPredictionTransformer<TModel> with the specified threshold. If the provided threshold is the same as the model threshold it simply returns model. Note that by default the threshold is 0.5 and valid scores range from 0 to 1.

Evaluate(IDataView, String, String, String, Int32)

Evaluates scored anomaly detection data.

Extension Methods

DetectEntireAnomalyBySrCnn(AnomalyDetectionCatalog, IDataView, String, String, SrCnnEntireAnomalyDetectorOptions)

Create Microsoft.ML.TimeSeries.SrCnnEntireAnomalyDetector, which detects timeseries anomalies for entire input using SRCNN algorithm.

DetectEntireAnomalyBySrCnn(AnomalyDetectionCatalog, IDataView, String, String, Double, Int32, Double, SrCnnDetectMode)

Create Microsoft.ML.TimeSeries.SrCnnEntireAnomalyDetector, which detects timeseries anomalies for entire input using SRCNN algorithm.

DetectSeasonality(AnomalyDetectionCatalog, IDataView, String, Int32, Double)

In time series data, seasonality (or periodicity) is the presence of variations that occur at specific regular intervals, such as weekly, monthly, or quarterly.

This method detects this predictable interval (or period) by adopting techniques of fourier analysis. Assuming the input values have the same time interval (e.g., sensor data collected at every second ordered by timestamps), this method takes a list of time-series data, and returns the regular period for the input seasonal data, if a predictable fluctuation or pattern can be found that recurs or repeats over this period throughout the input values.

Returns -1 if no such pattern is found, that is, the input values do not follow a seasonal fluctuation.

LocalizeRootCause(AnomalyDetectionCatalog, RootCauseLocalizationInput, Double, Double)

Create RootCause, which localizes root causes using decision tree algorithm.

LocalizeRootCauses(AnomalyDetectionCatalog, RootCauseLocalizationInput, Double, Double)

Outputs an ordered list of RootCauses. The order corresponds to which prepared cause is most likely to be the root cause.

Applies to