Crate cogset
Clustering algorithms.
This crate provides generic implementations of clustering
algorithms, allowing them to work with any back-end "point
database" that implements the required operations, e.g. one might
be happy with using the naive collection BruteScan from this
crate, or go all out and implement a specialised R*-tree for
optimised performance.
Density-based clustering algorithms:
- DBSCAN (
Dbscan) - OPTICS (
Optics)
Others:
- k-means (
Kmeans)
Installation
Add the following to your Cargo.toml file:
[dependencies]
cogset = "0.2"
Structs
| BruteScan | A point collection where queries are answered via brute-force scans over the whole list. |
| BruteScanNeighbours | An iterator over the neighbours of a point in a |
| Dbscan | Clustering via the DBSCAN algorithm[1]. |
| Euclid | Points in ℝn with the L2 norm. |
| Kmeans | Clustering via the k-means algorithm (aka Lloyd's algorithm). |
| KmeansBuilder | A builder for k-means to provide control over parameters for the algorithm. |
| Optics | Clustering via the OPTICS algorithm[1]. |
| OpticsDbscanClustering | An iterator over clusters generated by OPTICS using a DBSCAN-like criterion for clustering. |
Traits
| Euclidean | |
| ListPoints | Collections of points that can list everything they contain. |
| Point | A point in some (metric) space. |
| Points | A data structure that contains points of some sort. |
| RegionQuery | Collections of points that can be queried to find nearby points. |