enum Cadmium::Classifier::Tabular::DistanceMetric

Overview

Distance metrics for calculating similarity between feature vectors.

Used primarily by KNN to find nearest neighbors.

Defined in:

Enum Members

Euclidean = 0

Euclidean distance: √Σ(aᵢ - bᵢ)² Most common distance metric, works well for most cases

Manhattan = 1

Manhattan distance: Σ|aᵢ - bᵢ| Also known as L1 distance or city block distance Less sensitive to outliers than Euclidean

Chebyshev = 2

Chebyshev distance: max|aᵢ - bᵢ| Also known as L∞ distance or chessboard distance Useful for grid-like data

Cosine = 3

Cosine distance: 1 - (a·b)/(||a||·||b||) Measures angular similarity, ignores magnitude Useful for high-dimensional data

Instance Method Summary

• #chebyshev?

  Returns true if this enum value equals Chebyshev

• #cosine?

  Returns true if this enum value equals Cosine

• #euclidean?

  Returns true if this enum value equals Euclidean

• #manhattan?

  Returns true if this enum value equals Manhattan

Instance Method Detail