Module vq 

Vector quantization functions

This module provides vector quantization algorithms like K-means clustering and related utilities.

Examples

use scirs2_core::ndarray::{ArrayView1, Array2, ArrayView2};
use scirs2_cluster::vq::kmeans;

// Example data
let data = Array2::from_shape_vec((6, 2), vec![
    1.0, 2.0,
    1.2, 1.8,
    0.8, 1.9,
    3.7, 4.2,
    3.9, 3.9,
    4.2, 4.1,
]).unwrap();

// Run k-means with k=2
let (centroids, labels) = kmeans(ArrayView2::from(&data), 2, None, None, None, None).unwrap();

// Print the results
println!("Centroids: {:?}", centroids);
println!("Labels: {:?}", labels);

Structs

ChebyshevDistance

Chebyshev distance metric (L∞ norm)

CorrelationDistance

Correlation distance metric (1 - Pearson correlation)

CosineDistance

Cosine distance metric (1 - cosine similarity)

EuclideanDistance

Euclidean distance metric (L2 norm)

KMeansOptions

Options for K-means clustering

MahalanobisDistance

Mahalanobis distance metric using precomputed inverse covariance matrix

ManhattanDistance

Manhattan distance metric (L1 norm)

MiniBatchKMeansOptions

Options for Mini-Batch K-means clustering

MinkowskiDistance

Minkowski distance metric with configurable power p

ParallelKMeansOptions

Options for parallel K-means clustering

SimdOptimizationConfig

Configuration for SIMD optimizations

WeightedKMeansOptions

Options for weighted K-means clustering

Enums

KMeansInit

Initialization methods for K-means

MetricType

Enumeration of available distance metrics

MinitMethod

Initialization methods for kmeans2

MissingMethod

Methods for handling empty clusters during K-means clustering

Traits

VQDistanceMetric

Trait for distance metric computations

Functions

calculate_distortion_simd

SIMD-optimized distortion calculation

compute_centroids_simd

SIMD-optimized centroid computation for K-means

create_metric

Create a distance metric instance from the metric type

distance_to_centroids_simd

Compute distances from each point to a set of centroids using SIMD

euclidean_distance

Computes the Euclidean distance between two vectors

euclidean_distance_simd

SIMD-optimized Euclidean distance between two vectors

kmeans

K-means clustering algorithm (SciPy-compatible version)

kmeans2

Enhanced K-means clustering algorithm compatible with SciPy’s kmeans2

kmeans2_str

SciPy-compatible K-means clustering with string-based parameters

kmeans_init

K-means initialization algorithm

kmeans_plus_plus

K-means++ initialization algorithm

kmeans_plus_plus_simd

SIMD-accelerated K-means++ initialization

kmeans_simd

SIMD-accelerated K-means clustering

kmeans_with_metric

Enhanced K-means clustering with custom distance metrics

kmeans_with_options

K-means clustering algorithm (options-based version)

mini_batch_kmeans_simd

SIMD-accelerated mini-batch K-means for large datasets

minibatch_kmeans

Mini-Batch K-means clustering algorithm

pairwise_euclidean_parallel

Parallel distance matrix computation using core parallel operations

pairwise_euclidean_simd

Compute Euclidean distances between all pairs of points using SIMD when available

parallel_kmeans

Parallel K-means clustering algorithm

vq

Assign codes from a code book to observations.

vq_simd

SIMD-optimized vector quantization (assignment to nearest centroids)

weighted_kmeans

Weighted K-means clustering algorithm

weighted_kmeans_plus_plus

Weighted K-means++ initialization

whiten

Normalize a group of observations on a per feature basis.

whiten_simd

SIMD-optimized data whitening (normalization by standard deviation)