IQRDetector<T>

Models & Types · AiDotNet.AnomalyDetection.Statistical

Detects anomalies using the Interquartile Range (IQR) method.

Inheritance: ModelBase<T, Matrix<T>, Vector<T>> → AnomalyDetectorBase<T> → IQRDetector<T>

Implements: IGeneratedParameterRegistrar<T>

For Beginners

The IQR method is based on quartiles, which divide your data into four parts:

  • Q1 (25th percentile): 25% of data is below this value
  • Q3 (75th percentile): 75% of data is below this value
  • IQR = Q3 - Q1: The range containing the middle 50% of data

Outliers are points below Q1 - kIQR or above Q3 + kIQR, where k is the multiplier (typically 1.5).

How It Works

When to use:

  • Your data may not be normally distributed
  • You want a robust method that isn’t affected by extreme outliers
  • This is the same method used in box plots

Industry Standard Defaults:

  • Multiplier: 1.5 (identifies “mild” outliers)
  • Use 3.0 for “extreme” outliers only

Constructors

Constructor Summary
IQRDetector(double multiplier, double contamination, int randomSeed) Creates a new IQR-based anomaly detector.

Properties

Property Summary
Vector<T> IQR { get; } Gets the IQR values for each feature.
Vector<T> LowerBounds { get; } Gets the lower bounds (Q1 - multiplier * IQR) for each feature.
Vector<T> Q1 { get; } Gets the Q1 (25th percentile) values for each feature.
Vector<T> Q3 { get; } Gets the Q3 (75th percentile) values for each feature.
Vector<T> UpperBounds { get; } Gets the upper bounds (Q3 + multiplier * IQR) for each feature.
double Multiplier { get; } Gets the IQR multiplier for determining outlier boundaries.

Methods

Method Summary
Vector<T> ScoreAnomalies(Matrix<T> X)
void Fit(Matrix<T> X)