fastmath.clustering — generateme/fastmath 1.5.0-alpha4

Clustering algorithms.

Various clustering algrorithms backed by SMILE library.

Currently implemented: only partition clustering.

Input data

It's always sequence of n-sized samples as sequences.

For example, 2d samples [[1 2] [2 2] [3 3] ...]

For 1d data you can pass sequence of numbers of sequence of 1d seqs of numbers

[1 2 3]
;; or
[[1] [2] [3]]

Distances

Some of the methods use distance functions, use fastmath.distance namespace to create one.

Output

Every function returns record which contains:

• :type - name of the method used
• :data - input data
• :clustering - sequence of cluster ids
• :sizes - sizes of clusters
• :clusters - number of clusters
• :predict - predicting function (see below), qualify additional sample
• :representatives - list of centroids or medoids if available
• :info - additional statistics for your samples (like distortion)
• :obj - SMILE object

Cluster id is a integer ranging from 0 to the number of clusters minus 1. Some methods mark outliers with outlier-id.

Record acts as function and can qualify additonal sample by calling :predict function (or just call predict), for example (data is sequence of 3d samples):

(let [cl (k-means data 10)] (cl [0 1 2]))

See k-means

Regrouping

Clustering record can be regroupped to the list of individual clusters. Call regroup and get list of maps with following structure:

• :key - cluster id
• :data - samples which belong to the cluster
• :outliers? - does it contain outliers or not
• :representative - centroid/medoid or average vector if the former is not available
• :size - size of cluster

clarans clj

(clarans data clusters)

(clarans data dist clusters)

(clarans data dist clusters max-neighbor)

(clarans data dist clusters max-neighbor num-local)

Clustering Large Applications based upon RANdomized Search algorithm.

Input:

• data - sequence of samples
• clusters - numbe of clusters

Optional:

• dist - distance method, default :euclidean
• max-neighbor - maximum number of neighbors checked during random search
• num-local - the number of local minima to search for

See more in SMILE doc

clustering-methods-list clj

List of clustering methods.

dbscan clj

(dbscan data min-pts radius)

(dbscan data dist min-pts radius)

Density-Based Spatial Clustering of Applications with Noise algorithm.

Input:

• data - sequence of samples
• dist (optional) - distance method, default :euclidean
• min-pts - minimum number of neighbors
• radius - the neighborhood radius

See more in SMILE doc

denclue clj

(denclue data sigma m)

DENsity CLUstering algorithm.

Input:

• data - sequence of samples
• sigma - gaussian kernel parameter
• m - number of selected samples, much slower than number of all samples

See more in SMILE doc

deterministic-annealing clj

(deterministic-annealing data max-clusters)

(deterministic-annealing data max-clusters alpha)

Deterministic Annealing algorithm.

Input:

• data - sequence of samples
• max-clusters - number of clusters
• alpha (optional) - temperature decreasing factor (valued from 0 to 1)

See more in SMILE doc

g-means clj

(g-means data max-clusters)

G-Means algorithm.

Input:

• data - sequence of samples
• max-clusters - maximum number of clusters

See more in SMILE doc

k-means clj

(k-means data clusters)

(k-means data clusters max-iter)

(k-means data clusters max-iter runs)

K-Means++ algorithm.

Input:

• data - sequence of samples
• clusters - number of clusters
• max-iter (optional) - maximum number of iterations
• runs (optional) - maximum number of runs

See more in SMILE doc

mec clj

(mec data max-clusters radius)

(mec data dist max-clusters radius)

Nonparametric Minimum Conditional Entropy Clustering algorithm.

Input:

• data - sequence of samples
• dist (optional) - distance method, default :euclidean
• max-clusters - maximum number of clusters
• radius - the neighborhood radius

See more in SMILE doc

neural-gas clj

(neural-gas data clusters)

(neural-gas data clusters lambda-i lambda-f eps-i eps-f steps)

Neural Gas algorithm.

Input:

• data - sequence of samples
• clusters - number of clusters

Optional:

• lambda-i - intial lambda value (soft learning radius/rate)
• lambda-f - final lambda value
• eps-i - initial epsilon value (learning rate)
• eps-f - final epsilon value
• steps - number of iterations

See more in SMILE doc

outlier-id clj

Id of the cluster which contain outliers.

predict clj

(predict cluster in)

Predict cluster for given vector

regroup clj

(regroup clustered-data)

Transform clustering result into list of clusters as separate maps.

Every map contain:

• :key - cluster id
• :data - samples which belong to the cluster
• :outliers? - does it contain outliers or not
• :representative - centroid/medoid or average vector if the former is not available
• :size - size of cluster

Representative is always a n-dimensional sequence even if input is a list of numbers.

Empty clusters are skipped.

x-means clj

(x-means data max-clusters)

X-Means algorithm.

Input:

• data - sequence of samples
• max-clusters - number of clusters

See more in SMILE doc