core.coclustering_results¶

Submodule of khiops.core

Classes to access Khiops Coclustering JSON reports

Class Overview¶

Below we describe with diagrams the relationships of the classes in this modules. They are mostly compositions (has-a relations) and we omit native attributes (str, int, float, etc).

The main class of this module is CoclusteringResults and it is largely a composition of sub-reports objects given by the following structure:

CoclusteringResults
|- coclustering_report -> CoclusteringReport

CoclusteringReport
|- dimensions -> list of CoclusteringDimension
|- cells      -> list of CoclusteringCell

CoclusteringDimension
|- parts                    -> list of CoclusteringDimensionPart
|- variable_part_dimensions -> list of CoclusteringDimension
|- clusters                 -> list of CoclusteringCluster
|- root_cluster             -> CoclusteringCluster

CoclusteringDimensionPartValueGroup
|- values -> list of CoclusteringDimensionPartValue

CoclusteringCluster
|- leaf_part        -> CoclusteringDimensionPart or None
|- parent_cluster  |
|- child_cluster1  |-> CoclusteringCluster or None
|- child_cluster2  |

To have a complete illustration of the access to the information of all classes in this module look at their to_dict methods which write Python dictionaries in the same format as the Khiops JSON reports.

Functions¶

read_coclustering_results_file

Reads a Khiops Coclustering JSON report

Classes¶

`CoclusteringCell`	A coclustering cell
`CoclusteringCluster`	A cluster in a coclustering dimension hierarchy
`CoclusteringDimension`	A coclustering dimension (variable)
`CoclusteringDimensionPart`	An element of a partition of a dimension
`CoclusteringDimensionPartInterval`	An interval of a numerical partition
`CoclusteringDimensionPartValue`	A specific value of a variable in a dimension value group.
`CoclusteringDimensionPartValueGroup`	A value group of a categorical partition
`CoclusteringReport`	Main coclustering report
`CoclusteringResults`	Main class containing the information of a Khiops Coclustering JSON file

class khiops.core.coclustering_results.CoclusteringCell¶

Bases: object

A coclustering cell

Note

This class has only a no-parameter constructor initializing an instance with the default values.

Attributes:

partslist of CoclusteringDimensionPart: Parts for each coclustering dimension.
frequencyint: Frequency of this cell.

write_line(writer)¶

Writes a line of the instance’s report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12.

Parameters:

writerKhiopsOutputWriter: Output writer.

class khiops.core.coclustering_results.CoclusteringCluster(json_data=None)¶

Bases: object

A cluster in a coclustering dimension hierarchy

Parameters:

json_datadict, optional: JSON data of an element of the list at the dimensionHierarchies field within the coclusteringReport field of a Khiops Coclustering JSON report file. If not specified it returns an empty instance.

Attributes:

namestr: Name of the cluster.
parent_cluster_namestr: Name of the parent cluster.
frequencyint: Number of individuals in the cluster.
interestfloat: The cluster’s interest/informativeness.
hierarchical_levelfloat: A measure interpretable as the distance of the cluster to the root. Between 0 and 1.
rankint: Rank of clusters in the top-down list of clusters, with the smallest ranks at the top.
hierarchical_rankint: Rank of clusters in the hierarchy, with the smallest ranks being the closest from the root of the hierarchy.
is_leafbool: True if the cluster is a leaf of the hierarchy.
short_descriptionstr: Succinct cluster description.
descriptionstr: Cluster description.
leaf_partCoclusteringDimensionPart: On a leaf cluster: Its unique associated partition element. Otherwise None.
parent_clusterCoclusteringCluster: On a non-root cluster: Its unique parent cluster. Otherwise None.
child_cluster1CoclusteringCluster: On a non-leaf cluster : The first child cluster. Otherwise None.
child_cluster2CoclusteringCluster: On a non-leaf cluster : The second child cluster. Otherwise None.

to_dict()¶: Transforms this instance to a dict with the Khiops JSON file structure

write_annotation_header_line(writer)¶

Writes the “annotation” section’s header to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_annotation_line(writer)¶

Writes a line of the “annotation” section to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_hierarchy_header_line(writer)¶

Writes the “hierarchy” section’s header to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_hierarchy_line(writer)¶

Writes a line of the “hierarchy” section to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_hierarchy_structure_report(writer)¶

Writes the hierarchical structure from this instance to a writer object

This method is mainly a test of the encoding of the cluster hierarchy.

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

class khiops.core.coclustering_results.CoclusteringDimension¶

Bases: object

A coclustering dimension (variable)

A coclustering dimension is a hierarchical clustering of an input variable. The leaves of this hierarchy are linked to an element of a partition of the input variable. Leaf clusters have variable parts as their children.

It only has a no-parameter constructor.

Note

The instance information is initialized with the init_summary, init_partition and init_hierarchy methods. Its owner object (class CoclusteringReport) uses the information found in the fields dimensionSummaries, dimensionPartitions and dimensionHierarchies to coherently initialize the all dimensions with these methods.

Attributes:

namestr: Name of the variable associated to this dimension.
is_variable_partbool: True if the dimension is a part of a variable in an instance-variable coclustering.
type“Numerical” or “Categorical”: Dimension type.
part_numberint: Number of parts of the variable associated to this dimension.
initial_part_numberint: Number of initial parts. Note that part_number <= initial_part_number after a coclustering simplification (see simplify_coclustering).
value_numberint: Number of values of the dimension’s variable.
interestfloat: Interest of the dimension with respect to the other coclustering dimensions.
descriptionstr: Description of the dimension/variable.
minfloat: Minimum value of a numerical dimension/variable.
maxfloat: Maximum value of a numerical dimension/variable.
partslist of CoclusteringDimensionPart: Partition of this dimension.
variable_part_dimensionslist of CoclusteringDimension: Variable part instance-variable coclustering dimensions. None for variable-variable clustering.
clusterslist of CoclusteringCluster: Clusters of this dimension’s hierarchy. Note that includes intermediary clusters.
root_clusterCoclusteringCluster: Root cluster of the hierarchy.

get_cluster(cluster_name)¶

Returns the specified cluster

Parameters:

cluster_namestr: Name of the cluster.

Returns:

CoclusteringCluster: The specified cluster.

Raises:

KeyError: If there is no cluster with the specified name.

get_part(part_name)¶

Returns a part of the dimension given the part’s name

Parameters:

part_namestr: Name of the part.

Returns:

CoclusteringDimensionPart: The part with the specified name.

Raises:

KeyError: If there is no part with the specified name.

init_hierarchy(json_data)¶

Initializes the hierarchy attributes from a Python JSON object

Parameters:

json_datadict, optional: Python dictionary representing the data of an element of the list found at the dimensionHierarchies field of a Khiops Coclustering JSON report file. If not specified it leaves the object as-is.

Returns:

self: A reference to the caller instance.

init_partition(json_data=None)¶

Initializes the partition attributes from a Python JSON object

Parameters:

json_datadict, optional: Python dictionary representing the data of an element of the list found at the dimensionPartitions field of a Khiops Coclustering JSON report file. If not specified it leaves the object as-is.

Returns:

self: A reference to the caller instance.

init_summary(json_data=None)¶

Initializes the summary attributes from a Python JSON object

Parameters:

json_datadict, optional: Dictionary representing the data of an element of the list found at the dimensionSummaries field of a Khiops Coclustering JSON report file. If not specified it leaves the object as-is.

Returns:

self: A reference to the caller instance.

needs_annotation_report()¶

Status about the annotation report

Returns:

bool: True if the “annotation” section is reported

to_dict(report_type)¶

Transforms this instance to a dict with the Khiops JSON file structure

Parameters:

report_typestr: Type of the report. Can be either one of “summary”, “dimension”, and “hierarchy”.

write_annotation(writer)¶

Writes the “annotation” section to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_composition(writer)¶

Writes the “composition” section to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_dimension_header_line(writer)¶

Writes the “dimensions” section header to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_dimension_line(writer)¶

Writes the “dimensions” section line to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_hierarchy(writer)¶

Writes the “hierarchy” section to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_hierarchy_structure_report_file(report_file_path)¶

Writes the hierarchical structure of the clusters to a file

This method is mainly a test of the encoding of the cluster hierarchy.

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

report_file_pathstr: Path of the output file.

class khiops.core.coclustering_results.CoclusteringDimensionPart(json_data=None)¶

Bases: object

An element of a partition of a dimension

Abstract class.

Parameters:

json_data: dict, optional: See child classes for specific information about this parameter.

Attributes:

cluster_namestr: Name of the cluster to which this part belongs.

class khiops.core.coclustering_results.CoclusteringDimensionPartInterval(json_data=None)¶

Bases: CoclusteringDimensionPart

An interval of a numerical partition

Parameters:

json_datadict, optional: Python dictionary representing an element of type “Numerical” of the list at the dimensionPartitions field of a Khiops Coclustering JSON report file. If not specified it returns an empty instance.

Attributes:

cluster_namestr: Name of the cluster containing this interval.
lower_boundfloat: Lower bound of the interval.
upper_boundfloat: Upper bound of the interval.
is_missingbool: True if the instance’s represent the missing values. In this case lower_bound and upper_bound are set to None.
is_left_openbool: True if the interval is unbounded below lower_bound may contain the minimum value of the training data.
is_right_openbool: True if the interval is unbounded above upper_bound may contain the maximum value of training data.

Raises:

KhiopsJSONError: If json_data does not contain a “cluster” key.

part_type()¶

Part type of this instance

Returns:

str: Only possible value: “Interval”.

to_dict()¶: Transforms this instance to a dict with the Khiops JSON file structure

class khiops.core.coclustering_results.CoclusteringDimensionPartValue¶

Bases: object

A specific value of a variable in a dimension value group.

Note

This class has only a no-parameter constructor initializing an instance with the default values.

Attributes:

valuestr: String representation of the value.
frequencyint: Number of individuals having this value.
typicalityfloat: Indicates how much the value is representative of the cluster. Ranges from 0 to 1, 1 being completely representative.

class khiops.core.coclustering_results.CoclusteringDimensionPartValueGroup(json_data=None)¶

Bases: CoclusteringDimensionPart

A value group of a categorical partition

Parameters:

json_datadict, optional: Python dictionary representing an element of type “Categorical” of the list at the dimensionPartitions field of a Khiops Coclustering JSON report file. If None it returns an empty instance.

Attributes:

cluster_namestr: Name of the cluster containing this group.
valueslist of CoclusteringDimensionPartValue: The singleton parts composing this group part.
is_default_partbool: True if the instance represents the “unknown values” group.

Raises:

KhiopsJSONError: If json_data does not contain a “cluster” key.

part_type()¶

Part type of this instance

Returns:

str: Only possible value: “Value group”.

to_dict()¶: Transforms this instance to a dict with the Khiops JSON file structure

class khiops.core.coclustering_results.CoclusteringReport(json_data=None)¶

Bases: object

Main coclustering report

A coclustering is an unsupervised data grid equipped with additional structures to ease its exploration. In particular, it is a piecewise constant density estimator of the data distribution. The additional structures are the following:

A cluster hierarchy for each dimension
Indicators (such as the interest) for each variable, part and value.

A coclustering consists of one to many variables (dimensions), where each variable is partitioned as:

Intervals in the numerical case
Individual values or value groups in the categorical case.

The cross-product of the partitions forms a multivariate partition of cells and their frequencies allow to estimate the multivariate density.

In case of an unsupervised data grid, the cells are described by their index on the variable partitions, together with their frequencies.

Parameters:

json_datadict, optional: JSON data of the coclusteringReport field of a Khiops Coclustering JSON report file. If not specified it returns an empty instance.

Attributes:

instance_numberint: Number of individuals in the learning data table.
cell_numberint: Number of coclustering cells.
null_costfloat: Cost of the null model.
levelfloat: Measure between 0 and 1 measuring the information gain over the null model.
initial_dimension_numberint: Initial number of dimensions. The number of dimensions (len(dimensions)) may be less than this quantity after a simplification (see simplify_coclustering).
frequency_variablestr: Name of the variable to be aggregated in the cells. By default is the number of individuals.
dictionarystr: Name dictionary from which the model was learned.
databasestr: Path of the main training data table file.
sample_percentagefloat: Percentage of instances used in training.
sampling_mode“Include sample” or “Exclude samples”: Sampling mode used to split the train and datasets.
selection_variablestr: Variable used to select instances for training.
selection_valuestr: Value of selection_variable to select instances for training.
dimensionslist of CoclusteringDimension: Coclustering dimensions (variable).
cellslist of CoclusteringCell: Coclustering cells.

get_dimension(dimension_name)¶

Returns the specified dimension

Parameters:

dimension_namestr: Name of the dimension (variable).

Returns:

CoclusteringDimension: The specified dimension.

Raises:

KeyError: If no dimension with the specified names exist.

get_dimension_names()¶

Returns the names of the available dimensions

Returns:

list of str: The names of the available dimensions.

to_dict()¶: Transforms this instance to a dict with the Khiops JSON file structure

write_annotations(writer)¶

Writes the dimensions’ “annotation” sections to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_bounds(writer)¶

Writes the “bounds” section of the TSV report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_cells(writer)¶

Writes the “cells” section of the TSV report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_coclustering_stats(writer)¶

Writes the “stats” section of the TSV report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_compositions(writer)¶

Writes the dimensions’ “composition” sections to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_dimensions(writer)¶

Writes the “dimensions” section of the TSV report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_hierarchies(writer)¶

Writes the dimension reports’ “hierarchy” sections to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output writer for the report file.

write_report(writer)¶

Writes the instance’s TSV report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

writerKhiopsOutputWriter: Output stream or writer.

class khiops.core.coclustering_results.CoclusteringResults(json_data=None)¶

Bases: KhiopsJSONObject

Main class containing the information of a Khiops Coclustering JSON file

Parameters:

json_datadict, optional: Python dictionary representing the data of a Khiops Coclustering JSON report file. If not specified it returns an empty instance.

Note

Prefer either the the read_coclustering_results_file function from the core API to obtain an instance of this class from a Khiops Coclustering JSON file.

Attributes:

toolstr: Name of the Khiops tool that generated the JSON file.
versionstr: Version of the Khiops tool that generated the JSON file.
coclustering_reportCoclusteringReport: Coclustering modeling report.

to_dict()¶: Transforms this instance to a dict with the Khiops JSON file structure

write_report(stream_or_writer)¶

Writes the instance’s TSV report to a writer object

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

stream_or_writerio.IOBase or KhiopsOutputWriter: Output stream or writer.

write_report_file(report_file_path)¶

Writes a TSV report file with the object’s information

Warning

This method is deprecated since Khiops 11.0.0 and will be removed in Khiops 12. Use the to_dict method instead.

Parameters:

report_file_pathstr: Path of the output TSV report file.

khiops.core.coclustering_results.read_coclustering_results_file(json_file_path)¶

Reads a Khiops Coclustering JSON report

Parameters:

json_file_pathstr: Path of the JSON report file.

Returns:

CoclusteringResults: An instance of CoclusteringResults containing the report’s information.