Enum KMeansAlgorithm 

#[non_exhaustive]
pub enum KMeansAlgorithm {
    Lloyd,
    Hamerly,
}

Specifies the algorithm used for the KMeans assignment step.

Both variants minimise the same objective and, given identical initial centroids, converge to the same result. They only differ in how the assignment step is computed. Select a variant via KMeansParams::algorithm.

This setting only applies to batch fit. The incremental Mini-Batch K-means path (fit_with) always uses Lloyd’s update, and configuring Hamerly alongside fit_with is rejected with KMeansParamsError::IncrementalHamerly.

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

Lloyd

Standard Lloyd’s algorithm (also known as the “naive” algorithm).

On every iteration, computes the distance from each observation to every centroid to determine the closest one. Simple and predictable; work per iteration is O(n_observations * n_clusters * n_features).

Default variant. Works with any Distance.

Hamerly

Hamerly’s accelerated algorithm.

Uses the triangle inequality together with per-observation upper/lower distance bounds to skip most distance computations once the algorithm has stabilised. Produces the same result as Lloyd’s algorithm given the same initial centroids, and is typically substantially faster for well-separated clusters with a moderate number of centroids. For heavily overlapping clusters or very large n_clusters the bookkeeping overhead can make Lloyd a better choice.

Because the bounds rely on the triangle inequality, the supplied distance function must be a true metric. L2Dist, L1Dist and LInfDist satisfy this.

Only supported in batch fit; not available for Mini-Batch fit_with.

Reference: https://cs.baylor.edu/~hamerly/papers/sdm_2010.pdf

Trait Implementations

impl Clone for KMeansAlgorithm

fn clone(&self) -> KMeansAlgorithm

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for KMeansAlgorithm

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for KMeansAlgorithm

fn eq(&self, other: &KMeansAlgorithm) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for KMeansAlgorithm

impl Eq for KMeansAlgorithm

impl StructuralPartialEq for KMeansAlgorithm