Cluster analysis

Ver. 1.3.1 (2024-08-21)

This module provides a template class for online cluster analysis.

class mlpro.oa.streams.tasks.clusteranalyzers.basics.ClusterAnalyzer(p_cls_cluster: type = <class 'mlpro.oa.streams.tasks.clusteranalyzers.clusters.basics.Cluster'>, p_cluster_limit: int = 0, p_name: str = None, p_range_max=1, p_ada: bool = True, p_duplicate_data: bool = False, p_visualize: bool = False, p_logging=True, **p_kwargs)

Bases: OAStreamTask

Base class for online cluster analysis. It raises an event when a cluster was added or removed.

Steps to implement a new algorithm are: - Create a new class and inherit from this base class - Specify all cluster properties provided/maintained by your algorithm in C_CLUSTER_PROPERTIES. - Implement method self._adapt() to update your cluster list on new instances - Implement method self._adapt_reverse() to update your cluster list on obsolete instances - New cluster: hand over self._cluster_properties.values() on instantiation

Parameters:

• p_cls_cluster – Cluster class (Class Cluster or a child class).

• p_cluster_limit (int) – Optional limit for clusters to be created. Default = 0 (no limit).

• p_name (str) – Optional name of the task. Default is None.

• p_range_max (int) – Maximum range of asynchonicity. See class Range. Default is Range.C_RANGE_PROCESS.

• p_ada (bool) – Boolean switch for adaptivitiy. Default = True.

• p_duplicate_data (bool) – If True, instances will be duplicated before processing. Default = False.

• p_visualize (bool) – Boolean switch for visualisation. Default = False.

• p_logging – Log level (see constants of class Log). Default: Log.C_LOG_ALL

• p_kwargs (dict) – Further optional named parameters.

C_RESULT_SCOPE_ALL

Result scope, that includes all clusters

Type:

int = 0

C_RESULT_SCOPE_NONZERO

Result scope, that includes just clusters with result values > 0

Type:

int = 1

C_RESULT_SCOPE_MAX

Result scope, that includes just the cluster with the highest result value.

Type:

int = 2

C_CLUSTER_PROPERTIES

List of cluster properties supported/maintained by the algorithm. These properties are handed over to each new cluster.

Type:

PropertyDefinitions

C_TYPE = 'Cluster Analyzer'

C_EVENT_CLUSTER_ADDED = 'CLUSTER_ADDED'

C_EVENT_CLUSTER_REMOVED = 'CLUSTER_REMOVED'

C_PLOT_ACTIVE: bool = True

C_PLOT_STANDALONE: bool = False

C_RESULT_SCOPE_ALL: int = 0

C_RESULT_SCOPE_NONZERO: int = 1

C_RESULT_SCOPE_MAX: int = 2

C_CLUSTER_PROPERTIES: List[Tuple[str, int, bool, type]] = []

align_cluster_properties(p_properties: List[Tuple[str, int, bool, type]]) → list

Aligns list of cluster properties with the given list. In particular, the maximum derivative order of numeric properties is aligned.

Parameters:

• p_properties (PropertyDefinitions) – List of properties to be aligned with.

• Returns

• list – List of unknown properties.

_run(p_inst: Dict[int, Tuple[str, Instance]])

Custom method that is called by method run().

Parameters:

p_inst (InstDict) – Instances to be processed.

new_cluster_allowed() → bool

Determines whether adding a new cluster is allowed.

Returns:

True, if adding a new cluster allowed. False otherwise.

Return type:

bool

get_cluster_cls()

get_clusters() → dict[Cluster]

This method returns the current list of clusters.

Returns:

dict_of_clusters – Current dictionary of clusters.

Return type:

dict[Cluster]

_get_next_cell_id() → int

_add_cluster(p_cluster: Cluster) → bool

Protected method to be used to add a new cluster. Please use as part of your algorithm. Please use method new_cluster_allowed() before adding a cluster.

Parameters:

p_cluster (Cluster) – Cluster object to be added.

_remove_cluster(p_cluster: Cluster)

Protected method to remove an existing cluster. Please use as part of your algorithm.

Parameters:

p_cluster (Cluster) – Cluster object to be added.

_get_cluster_relations(p_relation_type: int, p_inst: Instance, p_scope: int) → List[Tuple[int, float, object]]

Internal method to determine the relation of the given instance to each cluster as a value in percent. Currently supported relations are membership and influence.

See also: public methods get_cluster_memberships() and get_cluster influences()

Parameters:

• p_relation_type (int) – Possible values are 0 (cluster membership) and 1 (cluster influence)

• p_inst (Instance) – Instance to be evaluated.

• p_scope (int) – Scope of the result list. See class attributes C_RESULT_SCOPE_* for possible values.

Returns:

results – List of result items which are tuples of a cluster id, a relative result value in [0,1] and a reference to the cluster object.

Return type:

List[ResultItem]

get_cluster_memberships(p_inst: Instance, p_scope: int = 2) → List[Tuple[int, float, object]]

Method to determine the relative membership of the given instance to each cluster as a value in [0,1].

See also: method Cluster.get_membership().

Parameters:

• p_inst (Instance) – Instance to be evaluated.

• p_scope (int) – Scope of the result list. See class attributes C_RESULT_SCOPE_* for possible values. Default value is C_RESULT_SCOPE_MAX.

Returns:

List of membership items which are tuples of a cluster id, a relative membership value in [0,1], and a reference to the cluster object.

Return type:

List[ResultItem]

get_cluster_influences(p_inst: Instance, p_scope: int = 2) → List[Tuple[int, float, object]]

Method to determine the relative influence of the given instance to each cluster as a value in [0,1].

See also: method Cluster.get_influence().

Parameters:

• p_inst (Instance) – Instance to be evaluated.

• p_scope (int) – Scope of the result list. See class attributes C_RESULT_SCOPE_* for possible values. Default value is C_RESULT_SCOPE_MAX.

Returns:

List of influence items which are tuples of a cluster id, a relative influence value in [0,1], and a reference to the cluster object.

Return type:

List[ResultItem]

init_plot(p_figure: Figure = None, p_plot_settings: PlotSettings = None)

Initializes the plot functionalities of the class.

Parameters:

• p_figure (Matplotlib.figure.Figure, optional) – Optional MatPlotLib host figure, where the plot shall be embedded. The default is None.

• p_plot_settings (PlotSettings) – Optional plot settings. If None, the default view is plotted (see attribute C_PLOT_DEFAULT_VIEW).

update_plot(p_inst: Dict[int, Tuple[str, Instance]] = None, **p_kwargs)

Specialized definition of method update_plot() of class mlpro.bf.plot.Plottable.

Parameters:

• p_inst (InstDict) – Instances to be plotted.

• p_kwargs (dict) – Further optional plot parameters.

remove_plot(p_refresh: bool = True)

” Removes the plot and optionally refreshes the display.

Parameters:

p_refresh (bool = True) – On True the display is refreshed after removal

_renormalize(p_normalizer: Normalizer)

Internal renormalization of all clusters. See method OATask.renormalize_on_event() for further information.

Parameters:

p_normalizer (Normalizer) – Normalizer object to be applied on task-specific

Ver. 2.1.0 (2024-07-08)

This module provides a template class for clusters to be used in cluster analyzer algorithms.

class mlpro.oa.streams.tasks.clusteranalyzers.clusters.basics.Cluster(p_id: int, p_properties: List[Tuple[str, int, bool, type]] = [], p_visualize: bool = False, **p_kwargs)

Bases: Id, Properties, KWArgs

Universal template class for a cluster with any number of properties added by a cluster analyzer.

Parameters:

• p_id (ClusterId) – Unique cluster id.

• p_properties (PropertyDefinitions) – List of property definitions.

• p_visualize (bool) – Boolean switch for visualisation. Default = False.

• p_kwargs (dict) – Further parameters.

C_PLOT_ACTIVE: bool = True

C_PLOT_STANDALONE: bool = False

C_PLOT_VALID_VIEWS: list = ['2D', '3D', 'ND']

C_PLOT_DEFAULT_VIEW: str = 'ND'

C_CLUSTER_COLORS = ['brown', 'olive', 'orange', 'green', 'red', 'gray', 'purple', 'pink', 'cyan', 'blue']

set_plot_color(p_color)

get_membership(p_inst: Instance) → float

Custom method to compute a scalar membership value for the given instance.

Parameters:

p_inst (Instance) – Instance.

Returns:

A scalar value in [0,1] that determines the given instance’s membership in this cluster. A value of 0 means that the given instance is not a member of the cluster at all while a value of 1 confirms full membership.

Return type:

float

get_influence(p_inst: Instance) → float

Custom method to compute a scalar influence value for the given instance.

Parameters:

p_inst (Instance) – Instance.

Returns:

Scalar value >= 0 that determines the influence of the cluster on the specified instance. A value 0 means that the cluster has no influence on the instance at all.

Return type:

float

property color

Ver. 1.3.0 (2024-06-18)

This module provides templates for cluster analysis to be used in the context of online adaptivity.

class mlpro.oa.streams.tasks.clusteranalyzers.clusters.centroid.ClusterCentroid(p_id: int, p_properties: List[Tuple[str, int, bool, type]] = [], p_visualize: bool = False, **p_kwargs)

Bases: Cluster

Extended cluster class with a centroid.

centroid

Centroid object.

Type:

Centroid

C_PROPERTIES: PropertyDefinitions = [('centroid', 0, False, <class 'mlpro.oa.streams.tasks.clusteranalyzers.clusters.properties.centroid.Centroid'>)]

set_id(p_id)

Sets/generates a new id.

Parameters:

p_id – Optional external id. If None, a unique id is generated.

get_influence(p_inst: Instance) → float

Default strategy to determine the influence of a cluster on a specified instance based on the metric distance between the instance and the cluster centroid.