| Did you know ... | Search Documentation: |
| Pack logtalk -- logtalk-3.100.1/docs/handbook/_sources/libraries/hdbscan_clusterer.rst.txt |
.. _library_hdbscan_clusterer:
hdbscan_clusterer
Simplified HDBSCAN-style clusterer. It builds the mutual-reachability
graph, computes a minimum spanning tree, derives the single-linkage
hierarchy, condenses the hierarchy using minimum_cluster_size, and
selects clusters using eom or leaf selection. Supports
continuous attributes only.
The library implements the clusterer_protocol defined in the
clustering_protocols library. It provides predicates for learning a
clusterer from a dataset, assigning new instances to clusters, and
exporting the learned clusterer as a list of predicate clauses or to a
file.
Datasets are represented as objects implementing the
clustering_dataset_protocol protocol from the
clustering_protocols library.
Open the `../../apis/library_index.html#hdbscan_clusterer <../../apis/library_index.html#hdbscan_clusterer>`__ link in a web browser.
To load this library, load the loader.lgt file:
::
| ?- logtalk_load(hdbscan_clusterer(loader)).
To test this library predicates, load the tester.lgt file:
::
| ?- logtalk_load(hdbscan_clusterer(tester)).
To run the performance benchmark suite, load the
tester_performance.lgt file:
::
| ?- logtalk_load(hdbscan_clusterer(tester_performance)).
minimum_cluster_size, and selects
clusters using eom or leaf selection.eom and leaf
cluster selection.noise is returned.The following options can be passed to the learn/3 predicate:
minimum_points(MinimumPoints): Minimum neighborhood size used when
computing core distances and mutual reachability. Default is 2.minimum_cluster_size(MinimumClusterSize): Minimum number of points
required for an extracted cluster. Default is 2.cluster_selection_method(Method): Cluster extraction policy.
Options: eom (default) or leaf.distance_metric(Metric): Distance metric to use. Options:
euclidean (default) or manhattan.feature_scaling(FeatureScaling): Whether to standardize continuous
attributes before clustering. Options: on (default) or off.The learned clusterer is represented as a compound term with the functor chosen by the user when exporting the clusterer and arity 4. For example:
::
hdbscan_clusterer(Encoders, Clusters, Noise, Options)
Where:
Encoders: List of continuous attribute encoders storing attribute
name, mean, and scale.
Clusters: List of
cluster(Id, Points, MaxCoreDistance, Stability) terms in
cluster-id order.Noise: List of encoded training points classified as noise.Options: Effective training options used to learn the clusterer.