Crate kmedoids[][src]

Expand description

k-Medoids Clustering with the FasterPAM Algorithm

For details on the implemented FasterPAM algorithm, please see:

Erich Schubert, Peter J. Rousseeuw
Fast and Eager k-Medoids Clustering:
O(k) Runtime Improvement of the PAM, CLARA, and CLARANS Algorithms

Under review at Information Systems, Elsevier.

Erich Schubert, Peter J. Rousseeuw:
Faster k-Medoids Clustering: Improving the PAM, CLARA, and CLARANS Algorithms
In: 12th International Conference on Similarity Search and Applications (SISAP 2019), 171-187.

This is a port of the original Java code from ELKI to Rust. But it does not include all functionality in the original benchmarks.

If you use this in scientific work, please consider citing above articles.


Given a dissimilarity matrix of size 4 x 4, use:

let data = ndarray::arr2(&[[0,1,2,3],[1,0,4,5],[2,4,0,6],[3,5,6,0]]);
let mut meds = kmedoids::random_initialization(4, 2, &mut rand::thread_rng());
let (loss, assi, n_iter, n_swap) = kmedoids::fasterpam(&data, &mut meds, 100);
println!("Loss is: {}", loss);


pub use crate::arrayadapter::ArrayAdapter;
pub use crate::safeadd::SafeAdd;


Adapter trait for accessing different types of arrays.

Add without overflow, for both integer and float types.


Run the Alternating algorithm, a k-means-style alternate optimization.

Run the FasterPAM algorithm.

Run the FastPAM1 algorithm, which yields the same results as the original PAM.

Implementation of the original PAM algorithm (BUILD + SWAP)

Implementation of the original PAM BUILD algorithm.

Implementation of the original PAM SWAP algorithm (no BUILD).

Random initialization (requires the rand crate)