Package nltk :: Package cluster :: Module gaac :: Class GAAClusterer

type GAAClusterer

           object --+        
                    |        
         api.ClusterI --+    
                        |    
util.VectorSpaceClusterer --+
                            |
                           GAAClusterer

The Group Average Agglomerative starts with each of the N vectors as singleton clusters. It then iteratively merges pairs of clusters which have the closest centroids. This continues until there is only one cluster. The order of merges gives rise to a dendrogram: a tree with the earlier merges lower than later merges. The membership of a given number of clusters c, 1 <= c <= N, can be found by cutting the dendrogram at depth c.

This clusterer uses the cosine similarity metric only, which allows for efficient speed-up in the clustering process.

Instance Methods

[hide private]

__init__(self, num_clusters=1, normalise=True, svd_dimensions=None)

source code

cluster(self, vectors, assign_clusters=False, trace=False)
Assigns the vectors to clusters, learning the clustering parameters from the data.

source code

cluster_vectorspace(self, vectors, trace=False)
Finds the clusters using the given set of vectors.

source code

update_clusters(self, num_clusters)

source code

classify_vectorspace(self, vector)
Returns the index of the appropriate cluster for the vector.

source code

Dendrogram

dendrogram(self)
Returns: The dendrogram representing the current clustering

source code

num_clusters(self)
Returns the number of clusters.

source code

_average_similarity(self, v1, l1, v2, l2)

source code

__repr__(self)

source code

Inherited from util.VectorSpaceClusterer: classify, likelihood, likelihood_vectorspace, vector

Inherited from util.VectorSpaceClusterer (private): _normalise

Inherited from api.ClusterI: classification_probdist, cluster_name, cluster_names

Method Details

[hide private]

init(self, num_clusters=1, normalise=True, svd_dimensions=None)
(Constructor)

source code

Parameters:

normalise - should vectors be normalised to length 1
svd_dimensions - number of dimensions to use in reducing vector dimensionsionality with SVD

Overrides: util.VectorSpaceClusterer.__init__

(inherited documentation)