AWS::APS::AnomalyDetector RandomCutForestConfiguration

Configuration for the Random Cut Forest algorithm used for anomaly detection in time-series data.

Syntax

To declare this entity in your CloudFormation template, use the following syntax:

JSON

{
  "IgnoreNearExpectedFromAbove" : IgnoreNearExpected,
  "IgnoreNearExpectedFromBelow" : IgnoreNearExpected,
  "Query" : String,
  "SampleSize" : Integer,
  "ShingleSize" : Integer
}

YAML

  IgnoreNearExpectedFromAbove: 
    IgnoreNearExpected
  IgnoreNearExpectedFromBelow: 
    IgnoreNearExpected
  Query: String
  SampleSize: Integer
  ShingleSize: Integer

Properties

IgnoreNearExpectedFromAbove

Configuration for ignoring values that are near expected values from above during anomaly detection.

Required: No

Type: IgnoreNearExpected

Update requires: No interruption

IgnoreNearExpectedFromBelow

Configuration for ignoring values that are near expected values from below during anomaly detection.

Required: No

Type: IgnoreNearExpected

Update requires: No interruption

Query

The Prometheus query used to retrieve the time-series data for anomaly detection.

Important

Random Cut Forest queries must be wrapped by a supported PromQL aggregation operator. For more information, see Aggregation operators on the Prometheus docs website.

Supported PromQL aggregation operators: avg, count, group, max, min, quantile, stddev, stdvar, and sum.

Required: Yes

Type: String

Minimum: 1

Update requires: No interruption

SampleSize

The number of data points sampled from the input stream for the Random Cut Forest algorithm. The default number is 256 consecutive data points.

Required: No

Type: Integer

Minimum: 256

Maximum: 1024

Update requires: No interruption

ShingleSize

The number of consecutive data points used to create a shingle for the Random Cut Forest algorithm. The default number is 8 consecutive data points.

Required: No

Type: Integer

Minimum: 2

Maximum: 1024

Update requires: No interruption