datapro4j
The data processing library for Java
The programmer’s guide
Revision: 1
Please, cite this document as:
J.R. Romero, J.M. Luna, S.
Ventura (2012). datapro4j: the data processing library for Java. Dept. of
Computer Science and Numerical Analysis, University of
Córdoba (Spain). Available
for download from http://www.uco.es/grupos/kdis/datapro4j
Knowledge Discovery and Intelligent Systems
University of Córdoba,
Spain
http://www.uco.es/grupos/kdis July
2012
CONTACT INFO
José Raúl Romero, PhD
Dept. Computer Science and Numerical Analysis
University of Córdoba, Spain
Email: jrromero@uco.es
Web: http://www.jrromero.net/en
PARTICIPANTS (BY
ALPHABETICAL ORDER)
• de la Torre López, José. BSc. [JTL]
• Luna, José María, MSc. [JML]
• Orozco Borrego, Mario. BSc. [MOB]
• Ramírez Quesada, Aurora. MSc. [ARQ]
PROJECT HISTORY
|
Version
|
Date
|
Description
|
Participants
|
|
0.1
|
July 2011
|
Initial version. Intruder algorithms.
|
ARQ, JTL, JML, JRR
|
|
0.2
|
September 2011
|
Strategies and columns
|
MOB, JML, JRR
|
|
0.3
|
April 2012
|
Refactoring, performance improvements
and testing
|
ARQ, JML, JRR
|
|
0.4
|
Under development
|
Weka wrappers for preprocessing, association, clustering and classification
|
JRR
|
|
0.5
|
Under development
|
New dataset sources from relational databases and
noSQL databases
|
JRR
|
DOCUMENT HISTORY
|
Revision
|
Date
|
Description
|
Author
|
|
1
|
July 17, 2012
|
Initial version of
this document
|
JRR
|
TABLE
OF FIGURES 6
Introduction 8
Purpose 8
Scope 8
License 8
Overview 9
To-do list 9
Package
es::uco::kdis::datapro 10
Package es::uco::kdis::datapro::algorithm 11
Package es::uco::kdis::datapro::algorithm::base 12
Class DatasetStrategy 12
Package
es::uco::kdis::datapro::algorithm::intruder 16
Class AverageAttack 16
Class BandwagonAttack 18
Class DatasetStatistics 21
Class IntruderAttack 22
Class LoveHateAttack 27
Class RandomAttack 29
Class ReverseBandwagonAttack 31
Class SegmentAttack 32
Package es::uco::kdis::datapro::algorithm::preprocessing 35
Package es::uco::kdis::datapro::algorithm::preprocessing::
discretization 36
Class EqualFrequencyDiscretization 39
Class EqualWidthDiscretization 36
Class MDLPDiscretize 40
Package es::uco::kdis::datapro::algorithm::preprocessing::
instance 43
Class RemoveDuplicates 43
Class RemovePercentage 44
Package es::uco::kdis::datapro::algorithm::validation 48
Class KFolds 48
Package es::uco::kdis::datapro::dataset 51
Class Dataset 51
Class FileDataset 64
Class InstanceIterator 68
Interface IIterator 70
Package es::uco::kdis::datapro::dataset::Column 72
Class ColumnAbstraction 72
Class ColumnImpl 79
Enumeration ColumnType 85
Class BinaryColumn 87
Class BinaryColumnImpl 89
Class CategoricalColumnImpl 95
Class DateColumn 100
Class DateColumnImpl 102
Class IntegerColumn 105
Class IntegerColumnImpl 108
Class NominalColumn 110
Class NominalColumnImpl 111
Class NumericalColumn 115
Class NumericalColumnImpl 119
Class RangeColumn 123
Class RangeColumnImpl 125
Package
es::uco::kdis::datapro::dataset::Source 130
Class ArffDataset 130
Class CsvDataset 135
Class ExcelDataset 139
Class KeelDataset 142
Package
es::uco::kdis::datapro::datatypes 146
Class InvalidValue 146
Class EmptyValue 147
Class MissingValue 147
Class NullValue 148
Class Range 149
Class DoubleRange 152
Package
es::uco::kdis::datapro::exception 154
Class IllegalFormatSpecificationException 154
Class NoSuchCategoryException 155
Class NotAddedValueException 156
Appendix A: UML diagrams 157
Package es.uco.kdis.datapro.algorithm.base 157
Package
es.uco.kdis.datapro.algorithm.preprocessing 158
Package
es.uco.kdis.datapro.dataset columns 159
Package es.uco.kdis.datapro.dataset.Source 160
Appendix B: Extending the
library 162
Project structure 162
Code documentation 163
Coding recommendations 164
Package es.uco.kdis.datapro.algorithm_ 11
Package
es.uco.kdis.datapro.algorithm.base 12
Class DatasetStrategy 12
Package es.uco.kdis.datapro.algorithm.intruder_ 16
Package
es.uco.kdis.datapro.algorithm.preprocessing_ 35
Package
es.uco.kdis.datapro.algorithm.preprocessing.discretization_ 36
Class
EqualFrequencyDiscretization_ 39
Class
EqualWidthDiscretization_ 36
Class MDLPDiscretize 40
Package
es.uco.kdis.datapro.algorithm.preprocessing.instance 43
Class RemoveDuplicates 43
Class RemovePercentage 45
Package es.uco.kdis.datapro.algorithm.validation_ 48
Package
es.uco.kdis.datapro.dataset 51
Class Dataset 52
Class FileDataset 64
Class InstanceIterator_ 69
Interface IIterator_ 70
Package
es.uco.kdis.datapro.dataset.Column_ 72
Abstract class
ColumnAbstraction_ 73
Abstract class ColumnImpl 80
Enumeration ColumnType 86
Class BinaryColumn_ 87
Class BinaryColumnImpl 89
Class CategoricalColumn_ 92
Class CategoricalColumnImpl 95
Class DateColumn_ 101
Class DateColumnImpl 102
Class IntegerColumn_ 105
Class IntegerColumnImpl 108
Class NominalColumn_ 110
Class NominalColumnImpl 112
Class NumericalColumn_ 115
Class NumericalColumnImpl 119
Class RangeColumn_ 123
Class RangeColumnImpl 125
Package
es.uco.kdis.datapro.dataset.Source 130
Class ArffDataset 130
Class CsvDataset 135
Class ExcelDataset 139
Class KeelDataset 143
Package
es.uco.kdis.datapro.datatypes 146
Class InvalidValue 146
Class EmptyValue 147
Class MissingValue 148
Class NullValue 149
Class Range 150
Class DoubleRange 152
Package
es.uco.kdis.datapro.exception_ 154
Class
IllegalFormatSpecificationException_ 154
Class NoSuchCategoryException_ 155
Class NotAddedValueException_ 156
Class diagram: package
overview_ 157
Class diagram: package
es.uco.kdis.datapro.algorithm.base 157
Class diagram: Package
es.uco.kdis.datapro.algorithm.preprocessing_ 158
Class diagram: Package
es.uco.kdis.datapro.dataset.Column_ 159
Package
es.uco.kdis.datapro.dataset.Source 160
Class diagram: Package
es.uco.kdis.datapro.datatypes 161
Class diagram: Package
es.uco.kdis.datapro.exception_ 161
This document provides class, interface, and enumeration specification for the datapro4j library. The specification provides the details of the
types being modeled within the system.
The datapro4j library is conceived to provide fully support to the efficient handling of data sets from different sources and declaring different kind of data
types. This task often takes too long to the Java programmer, especially in certain domains, such as analytical analysis or data mining. Notice that this library is
not provided for a given application domain, just for those that require the handling of
structured data in Java from diverse data sources.
Therefore, datapro4j can be used in data mining for handling inputs or
preprocessing data, using both internal strategies (e.g. algorithms on
instances, discretization, etc.) or external tools (e.g. Weka or any other application).
It can be also used for handling outputs: for example, in migrating data to other
different formats, rearrange results from external tools or algorithms,
executing statistical tests, etc.
It is worth mentioning that datapro4j is conceived to be extended,
adding new algorithms, data formats, column types, etc. All these aspects are
independent of each other, so algorithms can be executed regardless of being
introduced in diverse formats (stored in noSQL databases, as an ARFF file, or
whichever).
Scope
This document is intended to define the class specification for the datapro4j library.
Copyright Š 2012 The authors (University of Cordoba, Spain)
This software was developed by
members of the Knowlegde Discovery and Intelligent Systems at the University of Córdoba, Spain. For further information on the library and modifications, please
visit the URL http://www.uco.es/grupos/kdis/datapro4j
THE SOFTWARE IS PROVIDED "AS IS",
WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED.
Redistribution and use of
binary forms, with or without modification, are permitted if the following conditions are met:
ˇ
Redistributions of source code must retain the above copyright notice, this list of conditions and the
disclaimer above.
ˇ
Redistributions in
binary form must reproduce the above copyright notice, this list
of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
ˇ
All advertising materials or publication mentioning features or use of
this software must display the
following acknowledgement: “This product includes software developed by the KDIS Research
Group at the University of Córdoba (Spain) and its contributors.”
or cite the following reference:
J.R. Romero, J.M. Luna, S. Ventura (2012). datapro4j: the data processing library for Java. Dept. of Computer
Science and Numerical Analysis, University of Córdoba (Spain). Available
for download from http://www.uco.es/grupos/kdis/datapro4j
ˇ
Neither the name of
the University nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
ˇ
Commercial use of
this software or part of it is
not allowed without specific prior written permission.
ˇ
Licensing and conditions are subject to change without notice.
Note: At the moment this software is
provided in
binary form
as a Java library. Source code is
not provided
(we plan to release the Java source code in a near future).
Overview
This document provides a list of
all
packages with a summary for each. Each package has a section that contains a list of
its
classes, interfaces and enumeration type, with a summary for each. Class and Interface contains description, summary tables, detailed member descriptions, and relation table.
Private properties are omitted. Protected properties are shown when useful for external programmers.
To-do list
In the near future, this library will be updated with the following features (not necessarily in this
order):
ˇ
Listeners in strategies.
ˇ
Graphical UI. (Some minor support is already provided).
ˇ
Generation of synthetic datasets under precise constraints.
ˇ
Multipart datasets: those datasets which are not possible to be fully stored in memory, so they need
to be split and partially retrieved.
ˇ
Different data mining support.
ˇ
Wrappers for different datasets and tools.
o
A wrapper for Weka is
under development.
ˇ
Access to different databases.
o
Access thru JDBC to RDBMS engines (e.g. MySQL, Oracle) is
under development.
o
Access to no-sql engines (e.g. Cassandra) is
under development.
ˇ
More dataset formats:
o
Currently, the following formats are supported: ARFF, KEEL, CSV, Excel
o
The following formats are under development:
XRFF
The library base package. The software is
mainly divided into three different components:
ˇ
Dataset and columns. The logical abstract representation of a dataset and its attributes.
ˇ
Dataset and sources. The physical representation of a dataset, serialized in files, stored in databases or any other device.
ˇ
Dataset and strategies. Any algorithm running on a single dataset, set of datasets or column.
|
Name
|
datapro
|
|
Qualified Name
|
es::uco::kdis::datapro
|
Package es::uco::kdis::datapro::algorithm
Only those public strategies are described here. Developers can easily provide their own strategies.
Figure
1. Package es.uco.kdis.datapro.algorithm
|
Name
|
algorithm
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm
|
Package es::uco::kdis::datapro::algorithm::base
Figure
2. Package es.uco.kdis.datapro.algorithm.base
|
Name
|
base
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::base
|
Class
DatasetStrategy
This class represents a generic strategy.
Strategies are a well-known design pattern, where algorithms are encapsulated into classes. Strategies should be
executed using either a sequential or a step-by-step process. In
general, every strategy is executed according to the
following sequence:
ˇ
Creation:
the strategy constructor should collect all the parameters required by the algorithm to be initialized and executed for the first time. Build as many constructors as required.
ˇ
Initialization:
the method initialize() implements any preprocessing step required to
the algorithm to be executed. This preprocessing is not a part of
the algorithm itself but it should be executed for the first time that the algorithm is
invoked.
ˇ
Execution:
the method execution() runs the algorithm once using the parameters introduced when the constructor was invoked, and initialized afterwards. If the algorithm has finished and it could not be
invoked any more, then the method setExecutable(false) should be called. On the contrary, the execution is
allowed until the stop criteria are fulfilled.
ˇ
Stop criteria: the method isExecutable returns true if the algorithm can be executed once more over the dataset; false, otherwise.
ˇ
Post-execution: Any post-processing step has to be implemented by the method postexec().
ˇ
Result collection: Final results are collected from the dataset, if changed, and returned from the method getResult().
Figure
3. Class DatasetStrategy
|
Name
|
DatasetStrategy
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::base::DatasetStrategy
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Attribute Detail
bExecutable
Execution flag.
This is protected only for inheritance purposes, and should be never directly
modified.
|
Type
|
boolean
|
|
Default Value
|
true
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oDataset
Dataset used by the strategy.
|
Type
|
Dataset
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
execute
This method is invoked to execute the strategy.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
getDataset
Getter method for the dataset attribute.
|
Type
|
Dataset
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
This method returns an object comprising the resulting Object of the process
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
initialize
This method calls the Initialization process of the strategy.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
isExecutable
This method returns true if the strategy is in an executable state.
|
Type
|
boolean
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
postexec
This method should be invoked, if required, after the strategy execution.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
setDataset
This method sets the dataset to
be used by the strategy.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout data : Dataset
|
setExecutable
This method sets the current executable state of the
strategy.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bExecutable : boolean
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• EqualFrequencyDiscretization
|
|
Name
|
|
|
Related Element
|
• EqualWidthDiscretization
|
|
Name
|
|
|
Related Element
|
• MDLPDiscretize
|
|
Name
|
|
|
Related Element
|
• RemoveDuplicates
|
|
Name
|
|
|
Related Element
|
• IntruderAttack
|
|
Name
|
|
|
Related Element
|
• KFolds
|
|
Name
|
|
|
Related Element
|
• RemovePercentage
|
|
Name
|
|
|
Related Element
|
• DatasetStatistics
|
Package
es::uco::kdis::datapro::algorithm::intruder
Figure
4. Package
es.uco.kdis.datapro.algorithm.intruder
|
Name
|
intruder
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder
|
This class implements the Average Attack. This attack strategy sets the maximum value (push attack) or the
minimum value (nuke attack) to the target item. The filler items are selected randomly and their values are
also randomly chosen over a Normal Distribution, using the mean and standard deviation of the own item.
For a further description see the following paper:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. vol. 7, no. 4, pp.
23, 2007.
|
Name
|
AverageAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::AverageAttack
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• IntruderAttack
|
|
Realized Interface
|
|
Operation
Detail
AverageAttack
Parameterized Constructor.
• oDataset The original dataset
• iNumAttacks The number of attack instances
• bPush The attack type (true, push; false, nuke)
• iTarget The target item (The column attribute/item index)
• iNumFillers The size of filler item set: -1 for randomly size, >0 for fixed size
• dXRand The possibility of choose an
item as selected/filler item
• iSeed The random seed
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bPush : boolean
• in dXRand : double
• in iNumAttacks : int
• in iNumFillers : int
• in iSeed : int
• in iTarget : int
• inout oDataset : Dataset
|
chooseSelectedItems
The Average Attack does not use the selected item
set.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialization method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setFillerValues
In the Average Attack, the values for filler items must be randomly generated by a Normal Distribution, using the mean and standard deviation of each item.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setSelectedValues
The Average Attack does not use the selected item
set.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• IntruderAttack
|
Class
BandwagonAttack
This class implements the Bandwagon Attack. This attack strategy sets the maximum value (push attack) to
the target item. Then, a set of
items, named selected items, are chosen between the most visibility items.
The visibility items are those having a high mean and high evaluation density. For a further description see the following paper:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. vol. 7, no. 4, pp.
23, 2007.
|
Name
|
BandwagonAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::BandwagonAttack
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• IntruderAttack
|
|
Realized Interface
|
|
Attribute Detail
dDensity
The density threshold, i.e. the minimum number of values in the
column.
|
Type
|
double
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
dVisibility
The visibility threshold, i.e., the possibility of
choose an item to act as selected item.
|
Type
|
double
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
rgdMeanSD
It stores the mean and standard deviation of the
overall dataset.
|
Type
|
Double
|
|
Default Value
|
new ArrayList<Double>()
|
|
Visibility
|
protected
|
|
Multiplicity
|
0..*
|
rgoVisibilityColumns
The array of columns whose visibility exceed the thresholds dXVisibility and dXDensity.
|
Type
|
Integer
|
|
Default Value
|
new ArrayList<Integer>()
|
|
Visibility
|
package
|
|
Multiplicity
|
0..*
|
rgoVisibilityMeans
The array of mean columns whose visibility exceed the thresholds dXVisibility and dXDensity.
|
Type
|
Double
|
|
Default Value
|
new ArrayList<Double>()
|
|
Visibility
|
package
|
|
Multiplicity
|
0..*
|
Operation Detail
BandwagonAttack
Parameterized Constructor:
• oDataset The original dataset
• iNumAttacks The number of attack instances
• iTarget The target item (The column attribute/item index)
• iNumFillers The size of filler item set: -1 for randomly size, >0 for fixed size
• iNumSelected The size of
selected item set
• dVisibility The visibility threshold (absolute value of
column mean).
• dDensity The density threshold (absolute value of
instances without counting null, empty or
missing values in the column)
• dXRand The possibility of choose an
item as filler item
• iSeed The random seed
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dDensity : double
• in dVisibility : double
• in dXRand : double
• in iNumAttacks : int
• in iNumFillers : int
• in iNumSelected : int
• in iSeed : int
• in iTarget : int
• inout oDataset : Dataset
|
chooseSelectedItems
Create the set of selected items. The size is prefixed by iNumSelected property.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialization method for the strategy.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
orderArray
Order the columns using their mean as comparative metric. This method implements the QuickSort
algorithm.
• iInit The initial position of
the array
• iEnd The end position in the array
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iEnd : int
• in iInit : int
|
setFillerValues
In the Bandwagon Attack, the values for filler
items must be randomly generated by a Normal
Distribution, using the mean and standard deviation of the
overall dataset.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setSelectedValues
Set the values of selected items. In the Bandwagon Attack, each selected item
has
the maximum value.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setVisibilityColumns
Select the columns that exceed the visibility and density threshold.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ReverseBandwagonAttack
|
|
Name
|
|
|
Related Element
|
• IntruderAttack
|
Class
DatasetStatistics
|
Name
|
DatasetStatistics
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::DatasetStatistics
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
DatasetStatistics
Constructor. A parameter is required:
• data Dataset over which the statistical strategy will be executed.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout data : Dataset
|
execute
It executes the algorithm.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
It returns the mean and SD in form
of an ArrayList of Double values.
|
Type
|
ArrayList<Double>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Initialize
Inialization/Pre-processing method for the strategy.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
postexec
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
Class
IntruderAttack
IntruderAttack is the abstract base class for all the intruder attack algorithms. This class represents a generic attack used to
alter the content of
a dataset. It extends DatasetStrategy, whose methods
are implemented and adapted to a general intruder strategy.
For a further description see the paper:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. vol. 7, no. 4, pp.
23, 2007.
|
Name
|
IntruderAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::IntruderAttack
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
bPush
bPush represents the version of the
algorithm (true, for push attack; false for nuke attack).
|
Type
|
boolean
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
dXRand
dXrand represents the possibility of choosing an itemm(attribute)
as filler item.
|
Type
|
double
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
iActualInstance
iActualInstance represents the dataset instance modified by the attack.
|
Type
|
Int
|
|
Default Value
|
|
|
Visibility
|
Protected
|
|
Multiplicity
|
|
iNumAttacks
iNumAttacks represents the number of
attack instances that will be generated.
|
Type
|
int
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
iNumFillers
iNumFillers is the number of filler items, -1 if the filler item
set
size is randomly chosen.
|
Type
|
int
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
iNumSelected
iNumSelected is the number of selected items, -1 if the selected item
set
size is randomly chosen.
|
Type
|
Int
|
|
Default Value
|
|
|
Visibility
|
Protected
|
|
Multiplicity
|
|
iSeed
iSeed is the seed for the oRand object.
|
Type
|
Int
|
|
Default Value
|
|
|
Visibility
|
Protected
|
|
Multiplicity
|
|
iTarget
iTarget is the target attribute of the attack.
|
Type
|
int
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oInjection
oInjection stores the attack instances.
|
Type
|
Dataset
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oRand
oRand represents a random object.
|
Type
|
Random
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
rgoFillers
rgoFillers is the set of selected items.
|
Type
|
ColumnAbstraction
|
|
Default Value
|
new ArrayList<ColumnAbstraction>()
|
|
Visibility
|
protected
|
|
Multiplicity
|
0..*
|
rgoSelected
rgoSelected is the set of selected items.
|
Type
|
ColumnAbstraction
|
|
Default Value
|
new ArrayList<ColumnAbstraction>()
|
|
Visibility
|
protected
|
|
Multiplicity
|
0..*
|
Operation Detail
addAttack
Add a new instance (all items set to missed value) to the injection.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
chooseFillerItems
Select the set of
filler items. This set is common for all the intruder attack algorithms.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
chooseSelectedItems
Select the set of
selected items. The selection process is part of a specific intruder attack algorithm.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
true
|
|
Parameter
|
|
createRandomSetOfFiller
Select a random set of
columns to act as filler items. The set size is also randomly selected. It
returns the array of dataset columns that will act as filler
items.
|
Type
|
ArrayList<ColumnAbstraction>
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
createSetOfFiller
Select a random set of
columns to act as filler items. The set size is prefixed by iNumFiller property. It returns the array of
dataset columns that will act as filler items.
|
Type
|
ArrayList<ColumnAbstraction>
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
execute
Implements the strategy of attack algorithms.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMeanAndSD
Calculate the mean and standard deviation of the overall dataset. It returns an array with two elements, mean and standard deviation.
|
Type
|
ArrayList<Double>
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
Return the dataset injection created.
It returns the object comprising the injection after the attack.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialize the algorithm to prepare the execution.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
isSelectedColumn
This method returns a true value if the rgoSelected contains a column named as sName parameter,
false otherwise.
ˇ
sName The name of the column to be searched. It returns True if the column exists, false if not.
|
Type
|
boolean
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
ˇ
inout sName: String
ˇ
ˇ
|
postexec
Post-processing after the execute method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setFillerValues
This method assigns the correct
value for each filler item. It depends on the intruder attack algorithm.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
true
|
|
Parameter
|
|
setMaximumValue
Assign the maximum value to the target item.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setMinimumValue
Assign the minimum value to the target item.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setSelectedValues
The selected items value generation process. It is also depends on the specific intruder attack algorithm.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
true
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• AverageAttack
|
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
|
Name
|
|
|
Related Element
|
• RandomAttack
|
|
Name
|
|
|
Related Element
|
• LoveHateAttack
|
|
Name
|
|
|
Related Element
|
• BandwagonAttack
|
|
Name
|
|
|
Related Element
|
• SegmentAttack
|
Class
LoveHateAttack
This class implements the Love/Hate Attack. This attack strategy sets the maximum value (push attack) or the minimum value (nuke attack) to the target item. The filler items are selected randomly and their values are assigned in
the opposite sense of
the
target item.
For a further description see the paper:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. vol. 7, no. 4, pp.
23, 2007.
|
Name
|
LoveHateAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::LoveHateAttack
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• IntruderAttack
|
|
Realized Interface
|
|
Operation Detail
chooseSelectedItems
The Love/Hate Attack does not use the selected items.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialization method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
LoveHateAttack
Parameterized Constructor:
• oDataset The original dataset
• iNumAttacks The number of attack instances
• bPush The attack type (true, push; false, nuke)
• iTarget The target item (The column attribute/item index)
• iNumFillers The size of filler item set: -1 for randomly size, >0 for fixed size
• dXRand The possibility of choose an
item as selected/filler item
• iSeed The random seed
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in
bPush : boolean
• in dXRand : double
• in iNumAttacks : int
• in iNumFillers : int
• in iSeed : int
• in iTarget : int
• inout oDataset : Dataset
|
setFillerValues
In the Love/Hate Attack, the values for filler items must be
assigned in the opposite sense of
the
type of attack. If it is a push attack, all the filler items will be set to minimum value; if it is a nuke attack, all the
filler items will be set to maximum value.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setSelectedValues
The Love/Hate Attack does not use the selected items.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• IntruderAttack
|
Class
RandomAttack
This class implements the Random Attack. This attack strategy sets the maximum value (push attack) or the
minimum value (nuke attack) to the target item. The filler items are selected randomly and their values are
also chosen with a Normal Distribution, using the global dataset mean and standard deviation.
For a further description read the article:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. vol. 7, no. 4, pp.
23, 2007.
|
Name
|
RandomAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::RandomAttack
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• IntruderAttack
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
chooseSelectedItems
The Random Attack does not use the selected items.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialization method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
RandomAttack
Parameterized Constructor:
• oDataset The original dataset
• iNumAttacks The number of attack instances
• bPush The attack type (true, push; false, nuke)
• iTarget The target item (The column attribute/item index)
• iNumFillers The size of filler item set: -1 for randomly size, >0 for fixed size
• dXRand The possibility of choose an
item as selected/filler item
• iSeed The random seed
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bPush : boolean
• in dXRand : double
• in iNumAttacks : int
• in iNumFillers : int
• in iSeed : int
• in iTarget : int
• inout oDataset : Dataset
|
setFillerValues
In the Random Attack, the values for filler items must be randomly generated by a Normal Distribution,
using the mean and standard deviation of
the
dataset.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setSelectedValues
The Random Attack does not use the selected items.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• IntruderAttack
|
Class
ReverseBandwagonAttack
This class implements the Reverse Bandwagon Attack. This attack strategy sets the minimum value (nuke
attack) to the target item. Then, a set of
items, named selected items, are chosen between the less visibility
items. The visibility items are those having a low mean and high evaluation density.
For a better description read the article:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. vol. 7, no. 4, pp.
23, 2007.
|
Name
|
ReverseBandwagonAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::ReverseBandwagonAttack
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• BandwagonAttack
|
|
Realized Interface
|
|
Operation Detail
chooseSelectedItems
Create the set of selected items. The size is prefixed by iNumSelected property.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialization method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
ReverseBandwagonAttack
Parameterized Constructor:
• oDataset The original dataset
• iNumAttacks The number of attack instances
• iTarget The target item (The column attribute/item index)
• iNumFillers The size of filler item set: -1 for randomly size, >0 for fixed size
• iNumSelected The size of
selected item set: -1 for randomly size, >0 for fixed size
• dXVisibility The visibility threshold
• dXDensity The density threshold
• dXRand The possibility of choose an
item as selected/filler item
• iSeed The random seed
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dXDensity : double
• in dXRand : double
• in dXVisibility : double
• in iNumAttacks : int
• in iNumFillers : int
• in iNumSelected : int
• in iSeed : int
• in iTarget : int
• inout oDataset : Dataset
|
setSelectedValues
Set the values of selected items. In the Reverse Bandwagon Attack, each selected item
has the minimum value.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setVisibilityColumns
Select the columns that exceed the visibility and density threshold.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• BandwagonAttack
|
Class
SegmentAttack
This class implements the Segment Attack. This attack strategy sets the maximum value (push attack) to the target item. Then, a set of selected items (the segment) are set to
the maximum value. Finally, a set of
filler items are randomly chosen and the minimum value are set to
their.
For a better description read the article:
B. Mobasher, R. Burke, R. Bhaumik, and C. Williams. Toward Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithm Robustness. ACM Trans. Internet Technol. 7(4):1-23, 2007.
|
Name
|
SegmentAttack
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::intruder::SegmentAttack
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• IntruderAttack
|
|
Realized Interface
|
|
Attribute Detail
rgdMeanSD
rgdMeanSDstores the mean and standard deviation of the
overall dataset.
|
Type
|
Double
|
|
Default Value
|
new ArrayList<Double>()
|
|
Visibility
|
protected
|
|
Multiplicity
|
0..*
|
Operation Detail
chooseSelectedItems
Create the segment, the set of selected item, with the information given in rgsNamesOfSelected. It returns the array of
dataset columns that will act as selected items.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialization method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
SegmentAttack
Parameterized Constructor:
• oDataset The original dataset
• iNumAttacks The number of attack instances
• iTarget The target item (The column attribute/item index)
• iNumFillers The size of filler item set: -1 for randomly size, >0 for fixed size
• rgsNamesOfSelected The array with the names of the columns that will act as
selected items (the segment)
• dXRand The possibility of choose an
item as selected/filler item
• iSeed The random seed
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dXRand : double
• in iNumAttacks : int
• in iNumFillers : int
• in iSeed : int
• in iTarget : int
•
inout oDataset : Dataset
•
inout rgsNamesOfSelected : ArrayList<String>
|
setFillerValues
Set the value for filler
items. In the Segment Attack, the minimum value is assigned.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setSelectedValues
Set the values for the selected items. In the Segment Attack, the maximum value is assigned.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• IntruderAttack
|
Figure
6. Package
es.uco.kdis.datapro.algorithm.preprocessing.discretization
|
Name
|
discretization
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::discretization
|
Class
EqualWidthDiscretization
Equal-width discretization of
a given numerical/integer column of the
dataset. A RangeColumn is returned. Notice that this class is inherited from EqualFrequencyDiscretization.
Figure
8. Class EqualWidthDiscretization
|
Name
|
EqualWidthDiscretization
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::discretization::EqualWidthDi
scretization
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
iBins
iBins is the number of bins.
|
Type
|
int
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oCol
The column to be discretized.
|
Type
|
NumericalColumn
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oRangeColumn
The column returned as result.
|
Type
|
RangeColumn
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sColName
The name of
the column to be
discretized.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sResName
The name of
the resulting column.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
calculateDRangeColumn
This (protected) method creates a new RangeColumn taking both the intervals given as parameter and the
values comprised by the original numerical column.
• aoRanges Array of
intervals
• sName Name of the new column
It returns the
resulting RangeColumn.
|
Type
|
RangeColumn
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout aoRanges : DoubleRange
• inout sName : String
|
EqualWidthDiscretization
Parameterized Constructor:
• oDataset The dataset to be processed.
• iBins The number of bins.
• sColName The name of the column to be processed.
• sResName The name of the resulting column .
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iBins : int
• inout oDataset : Dataset
• inout sColName : String
• inout sResName : String
|
execute
This method
runs the discretization process. Firstly, it calculates the
cut-points and sets the range intervals.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
The discretized RangeColumn is
returned.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
The initialization method.
Types of the column and its
values are checked.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
postexec
Not required.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• EqualFrequencyDiscretization
|
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
Class
EqualFrequencyDiscretization
Equal-frequency discretization of a given numerical/integer column of the
dataset. A RangeColumn is returned.
Figure
7. Class EqualFrequencyDiscretization
|
Name
|
EqualFrequencyDiscretization
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::discretization::EqualFrequen
cyDiscretization
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
• EqualWidthDiscretization
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
Notice that this class is inherited from EqualWidthDiscretization.
EqualFrequencyDiscretization
Parametrized constructor.
Parameters:
ˇ
iBins Number of bins to be created
ˇ
oDataset Source dataset containing the column to be discretized
ˇ
sColName Name of the source column
ˇ
sResName Name of the resulting Range column
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iBins : int
• inout oDataset : Dataset
• inout sColName : String
• inout sResName : String
|
execute
This method makes the discretization by
frequency of the column passed as parameter.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
|
Name
|
|
|
Related Element
|
• EqualWidthDiscretization
|
Class
MDLPDiscretize
Figure
9. Class MDLPDiscretize
|
Name
|
MDLPDiscretize
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::discretization::MDLPDiscreti
ze
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
execute
This method runs the discretization process
following the MDLP algorithm.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
It returns the discretized dataset.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
The initialize() strategy method. It takes the whole dataset, and distribute each column in a LinkedList that contains a double array where the first value is
the
concrete value of the
column, the second value is
the label associated.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
MDLPDiscretize
Constructor with parameters:
• oDataset source dataset
Note: class labels are supposed to
be in the last column of
the dataset.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDataset : Dataset
|
postexec
The postexec() strategy method
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
Figure
10. Package
es.uco.kdis.datapro.algorithm.preprocessing.instance
|
Name
|
instance
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::instance
|
Class
RemoveDuplicates
This class modifies the content of
a Dataset by removing duplicate instances from
this dataset.
Figure
11. Class RemoveDuplicates
|
Name
|
RemoveDuplicates
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::instance::RemoveDuplicates
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
execute
Execution method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
It returns the clean dataset.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialize the algorithm to prepare the execution.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
postexec
Post-processing.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
RemoveDuplicates
Parameterized Constructor:
• oDataset The source dataset to
work
with.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDataset : Dataset
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
Class
RemovePercentage
This class modifies the content of
a dataset by removing a percentage of its instances.
Figure
12. Class RemovePercentage
|
Name
|
RemovePercentage
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::preprocessing::instance::RemovePercentag
e
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
RANDOM
RANDOM mode, when instances to be removed are randomly selected.
|
Type
|
int
|
|
Default Value
|
0
|
|
Visibility
|
public
|
|
Multiplicity
|
|
FROMINIT
FROMINIT mode, when instances to
be removed are taken from the beginning of the
column.
|
Type
|
int
|
|
Default Value
|
1
|
|
Visibility
|
public
|
|
Multiplicity
|
|
FROMEND
FROMEND mode, when instances to
be removed are taken from
the end of the column.
|
Type
|
int
|
|
Default Value
|
2
|
|
Visibility
|
public
|
|
Multiplicity
|
|
oRnd
oRnd is the random generator object.
|
Type
|
Random
|
|
Default Value
|
new Random()
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Operation Detail
execute
Execute method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
Return the resulting dataset from the strategy process.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
Initialize the algorithm to prepare the execution.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
postexec
Post-processing method.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
RemovePercentage
Parameterized Constructor:
• oDataset The source dataset
• iMode The mode of removal
• dPercentage The percentage of instances (in [0,1]) to remove from
the dataset
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dPercentage : double
• in iMode : int
• inout oDataset : Dataset
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
Figure
13. Package
es.uco.kdis.datapro.algorithm.validation
|
Name
|
validation
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::validation
|
This class implements the strategy that calculates the different partitions of the
dataset using the KFolds algorithm.
Figure
14. Class es.uco.kdis.datapro.algorithm.validation.KFolds
|
Name
|
KFolds
|
|
Qualified Name
|
es::uco::kdis::datapro::algorithm::validation::KFolds
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• DatasetStrategy
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
execute
It runs the KFolds algorithm. After the execution, the algorithm is
not executable anymore.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getResult
This method returns the list containing the resulting dataset partitions.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
initialize
This method initializes the algorithm. The instances are sorted as a HashMap by categories.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
KFolds
Parameterized constructor. Notice that the class column is
supposed to be the last column in the
dataset:
• oDataset Source dataset
• iNumberOfPartitions Number of partitions to
be built
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iNumberOfPartitions : int
• inout oDataset : Dataset
|
KFolds
Parameterized constructor. Notice that the class column is
supposed to be the last column in the
dataset:
• oDataset Source dataset
• iNumberOfPartitions Number of partitions to
be built
• iSeed If the programmer wants to reproduce a previous
partition, he can indicate a given seed to the process. Otherwise, the seed is
randomly selected.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iNumberOfPartitions : int
• inout oDataset : Dataset
|
postexec
Not required.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• DatasetStrategy
|
Package es::uco::kdis::datapro::dataset
Figure
15. Package es.uco.kdis.datapro.dataset
|
Name
|
dataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset
|
Dataset is
the
abstract base class for all the different types of
dataset sources. This class fills the gap
between the physical dataset (stored in a file, database, etc.) and its logical handling, where the access to attributes/columns and processing methods
is provided.
Figure
16. Class Dataset
|
Name
|
Dataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Dataset
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Attribute Detail
iCursor
iCursor refers to the row being pointed in the dataset by the InstanceIterator.
|
Type
|
int
|
|
Default Value
|
|
|
Visibility
|
Protected
|
|
Multiplicity
|
|
rgoColumns
rgoColumns is the list of columns that comprise the dataset.
|
Type
|
ColumnAbstraction
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
0..*
|
rgoValidBinaryFalseValues
For binary columns, it contains the list of
values that will be
interpreted as False when reading from the physical dataset. Writing will be performed using the first element in the
list.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
Protected
|
|
Multiplicity
|
0..*
|
rgoValidBinaryTrueValues
For binary columns, it contains the list of
values that will be
interpreted as True when reading from
the physical dataset. Writing will be performed using the first element in the
list.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
0..*
|
sOpenRangeDelimiter
For range columns, sOpenRangeDelimiter stores the symbol(s) that open the numerical range, right before the minimum value: e.g., '[' for [2,3]. This is used during the reading and writing of
the
physical dataset.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sSeparationRangeDelimiter
For range columns, sSeparationRangeDelimiter stores the symbol(s) that separate the minimum
and maximum values in a numerical range: e.g., ',' for [2,3]. This value is only used during the reading and
writing of
the physical dataset.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sCloseRangeDelimiter
For range columns, sCloseRangeDelimiter stores the symbol(s) that serves to close the numerical range, right after the maximum value: e.g., ']' for [2,3]. This is
only
used during the reading and writing of the physical dataset.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
tiplicity
sEmptyValue
sEmptyValue stores the string that will represent an empty value in the
dataset file.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sMissingValue
sMissedValue stores the string that will represent a missing value in the dataset file.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sNullValue
sNullValue stores the string that will represent a null value in
the dataset file.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sName
The name of
the dataset.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
addAllValues
A set of
column values are inserted into the dataset structure. Notice that instance duplication is not
checked.
Parameters:
• sColumnFormat String that specifies the types of the columns to be
added. Types depend
on the specific dataset.
Exceptions:
• IOException
• IllegalFormatSpecificationException
• NotAddedValueException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
true
|
|
Parameter
|
• inout sColumnFormat : String
|
addColumn
Insert a column abstraction given by parameter in the last position of the
list of columns of the dataset
Parameter:
• oColumn: Column abstraction to
be added
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oColumn : ColumnAbstraction
|
addColumn
Insert a column abstraction in a given position of
the
list of dataset columns.
Parameters:
• oColumn: Column abstraction to be inserted
• iIndex: Position index where the column element is
added
in the list. The rest of column items will be shifted one position to the right.
Exceptions:
• UnsupportedOperationException
• ClassCastException
• NullPointedException
• IllegalArgumentException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iIndex : int
• inout oColumn : ColumnAbstraction
|
clone
Create a new dataset exactly with the same metadata
and column structure. However, only the structure is
copied, since instances from
the original dataset are not added to
the new one.
It returns the empty cloned dataset.
|
Type
|
Dataset
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
close
Abstract method that serves to close the physical dataset source.
Exceptions:
• IOException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
true
|
|
Parameter
|
|
copy
This method creates a new dataset exactly with the same metadata,
column structure and data than the
original dataset. In this case, instances from the original dataset are also copied to the new one.
A copy of the
dataset is returned.
|
Type
|
Dataset
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Dataset
This is
the default constructor of this class. By default, it sets the following parameters to their default
values:
• sMissedValue: "?"
• sNullValue: "?"
• sEmptyValue: "?"
• sOpenRangeDelimiter: "["
• sSeparationRangeDelimiter: ","
• sCloseRangeDelimiter: "]"
Notice that using these symbols is not mandatory for reading/writing, as its applicability depends on the
specific implementation of
each
source dataset.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getColumn
This method looks for a column abstraction by its index in the column list. Notice that indexes can change when one column is
added
or removed to/from intermediate positions.
Parameter:
• iIndex: Index of the queried column.
It returns a reference to
the column abstraction queried.
|
Type
|
ColumnAbstraction
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
getColumnByName
This method returns the first column instance found having the name required as parameter.
Parameter:
• sName: The name of the column queried (no case-sensitive)
It returns the column abstraction class that accesses to the column required by its name.
|
Type
|
ColumnAbstraction
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
getColumns
Getter method for the private property rgoColumns, which comprises the array of column abstractions
in
the dataset.
|
Type
|
List<ColumnAbstraction>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getEmptyValue
Getter method for the private property sEmptyValue, which comprises the String that represents the symbol for the empty value in the dataset source. Notice that each specific dataset implementation can use, or not, this property accordingly.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getIndexOfColumn
Given a column abstraction, it searches for the index that this column occupies in
the array of column
abstractions in the
dataset.
Parameter:
• oCol: Column to be located.
It returns the index of the column abstraction passed as parameter; -1, otherwise.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oCol : ColumnAbstraction
|
getMissingValue
Getter method for the private property sMissingValue, which comprises the String that represents the
symbol for the missing value in the
dataset source. Notice that each specific dataset implementation
can
use, or not, this property accordingly.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getName
Getter method for the private property sName, which represents the name given to
the dataset.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getNullValue
Getter method for the private property sNullValue, which comprises the String that represents the
symbol for the null value in
the dataset source. Notice that each specific dataset implementation can use
or not this property accordingly.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getNumberOfDecimals
Getter method for the private property iNumberOfDecimals, which indicates the number of decimal
digits used when writing numerical columns in dataset sources. Notice that this value can be used
accordingly by each specific dataset source.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getRangeDelimiters
This method gets a list of
the
three values used to demarcate a range, comprising the sOpenRangeDelimiter, sSeparationRangeDelimiter and sCloseRangeDelimiter. Notice that each specific dataset source could make use of
these values accordingly.
|
Type
|
ArrayList<String>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValidBinaryFalseValues
Getter method for the private property rgoValidBinaryFalseValues: the list of strings that are interpreted as false when reading and writing Boolean values in a dataset source. Notice that its specific use depends on
the
implementation made for each dataset source.
|
Type
|
ArrayList<String>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValidBinaryTrueValues
Getter method for the private property rgoValidBinaryTrueValues: the list of strings that are
interpreted as true when reading and writing Boolean values in
a dataset source. Notice that its specific use depends on the implementation made for each dataset source.
|
Type
|
ArrayList<String>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
merge
This method merges two datasets by adding the dataset passed as parameter to the current one. Parameters:
• oDSInjected: The dataset to be added. Notice that this dataset must contain the same number and type of columns than the dataset object this.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDSInjected : Dataset
|
merge
This method merges two datasets by adding the dataset passed as parameter to the dataset object
this.
Parameters:
ˇ
oDataset: The dataset to be added.
ˇ
sColumnFormat: Sometimes the target dataset contains more columns than the source
dataset. For those cases, the columns to be added can be explicitly specified. This parameter is
a String that indicates the columns to be added. Each character in the String matches to a column in the target dataset. The String may comprise some of
the following characters:
o
x: Include this column
o
%: Skip this column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDataset : Dataset
• inout sColumnFormat : String
|
open
Abstract protected method. This method just opens the source dataset and initializes the row cursor to the first row of data. However, each specific dataset class is
responsible for its implementation, and thus
defining its real scope, according to
its
specific properties.
Notice that each type of datasets will provide specific methods to process the full dataset. For example,
file datasets provide the method readDataset.
Exceptions:
• FileNotFoundException
• IOException
• IllegalFormatSpecificationException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
true
|
|
Parameter
|
|
removeColumn
This method removes a column from the dataset. Notice that column indexes can be modified (decreased) for the rest of columns. The column removed is returned.
Parameter:
• iIndex: Position index where the column to be removed is located.
Exceptions:
• UnsupportedOperationException
• IndexOutOfBoundsException
|
Type
|
ColumnAbstraction
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
setColumns
Setter method for the property rgoColumns. Even when it is
a public method, notice that it should be used very carefully, mainly for those cases when the replacement of the entire set of columns is
mandatory. To add or remove a single column, or just a set of them, use instead the methods
addColumn and removeColumn.
Parameter:
• rgoCols: The entire list of columns in the dataset.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoCols : List<ColumnAbstraction>
|
setEmptyValue
Setter method for the private property sEmptyValue, which comprises the String that represents the symbol for the null value in the dataset source. Notice that each specific dataset implementation can make its own use of this property accordingly.
Parameters:
• sEmptyValue The symbol/string representing an empty value in the
dataset
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sEmptyValue : String
|
setMissingValue
Setter method for the private property sMissingValue, which comprises the String that represents the
symbol for the missing value in the
dataset source. Notice that each specific dataset implementation
can
make its own use of
this
property accordingly.
Parameters:
• sMissingValue The symbol/string representing a missing value in the dataset
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sMissingValue : String
|
setName
Setter method for the private property sName, which represents the name of the dataset. Parameter:
• sName: The name of
the dataset.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
setNullValue
Setter method for the private property sNullValue, which comprises the String that represents the
symbol for the null value in
the dataset source. Notice that each specific dataset implementation can
make its own use of
this property accordingly.
Parameters:
• sNullValue The symbol/string representing a null value in the dataset
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sNull : String
|
setNumberOfDecimals
Setter method for the private property iNumberOfDecimals, which represents the number of
decimals
that the programmer wants to set for numerical values. Notice that the specific applicability of this
attribute directly depends on the specific implementation of
the dataset source.
Parameter:
• iNum: The number of
decimal digits that will be considered when saving numerical values.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iNum : int
|
setRangeDelimiters
This method sets the symbols that will serve as range delimiter. Notice that the specific applicability of
these attributes directly depends on the specific implementation of
the
dataset source.
Parameters:
• sInitial: The symbol/string that represents the starting delimiter.
• sSeparator: The symbol/string that represents the value separator.
• sEnding: The symbol/string that represents the ending delimiter.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sEnding : String
• inout sInitial : String
• inout sSeparator : String
|
setValidBinaryFalseValues
Setter method of the
list rgoValidBinaryFalseValues, which contains the set of
strings that
represent a False boolean value in
the dataset. Notice that only the first value of
the list should be
used for serialization (reading or writing). In
any
case, it depends on the specific implementation of each dataset source.
Parameter:
• rgoValidBinaryFalseValues: The list of
values that will be interpreted as False.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoValidBinaryFalseValues :
ArrayList<String>
|
setValidBinaryTrueValues
Setter method of the
list rgoValidBinaryTrueValues, which contains the set of
strings that
represent a True boolean value in
the dataset. Notice that only the first value of the
list should be used
for
serialization (reading or writing). In any case, it depends on the specific implementation of each dataset source.
Parameter:
• rgoValidBinaryTrueValues: The list of
values that will be
interpreted as True.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoValidBinaryTrueValues
: ArrayList<String>
|
setValidBinaryValues
This method sets both the list of strings that will represent a
True boolean value, and the list
of strings that will represent a False boolean value in the dataset. This functionality could be also done by invoking seldom specific methods.
Parameters:
• rgoFalseList: A list with
the valid False symbols/strings
• rgoTrueList: A list with the valid True symbols/strings
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoFalseList : ArrayList<String>
• inout rgoTrueList : ArrayList<String>
|
swapColumns
This method swaps two columns in the list of columns of the
dataset. It searches for both columns, and
swaps its positions, and thus both structure and data.
Parameters:
• oColumn1: The first column to swap.
• oColumn2: The second column to swap.
Exceptions:
• ColumnAbstraction
• UnsupportedOperationException
• ClassCastException
• NullPointedException
• IllegalArgumentException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oColumn1 : ColumnAbstraction
• inout oColumn2 : ColumnAbstraction
|
Relation Detail
Association
|
Name
|
rgoColumns
|
|
Related Element
|
• ColumnAbstraction
|
Dependency
|
Name
|
|
|
Related Element
|
• InstanceIterator
|
Generalization
|
Name
|
|
|
Related Element
|
• FileDataset
|
Class
FileDataset
This abstract class represents a dataset when its source is extracted from a file. It includes the specific
methods required to
handle with datasets in form
of files.
Figure
17. Class FileDataset
|
Name
|
FileDataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::FileDataset
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
• Dataset
|
|
Realized Interface
|
|
Attribute Detail
oBufferedReader
oBufferedReader is the buffer used to read the file.
|
Type
|
BufferedReader
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sCommentValue
sCommentedValue stores the string that will indicate the beginning of
a comment line in the dataset
file, if this line has to
be omitted from the processing.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sFileName
sFileName is the name of the file source that contains the dataset.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
sSeparationSymbol
sSeparationSymbol stores the symbol/string that indicates the separator between values of the
same instance-row (i.e., a comma, a line of the dataset file, etc).
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
clone
This method creates a new dataset exactly with the same type and column structure than the original.
Instances from
the original dataset are not copied. It returns a new Dataset instance.
|
Type
|
Dataset
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
copy
This method clones the dataset and fills its content with the instances extracted from the original. Create a new dataset exactly with the same type, column structure and data. It returns the copied Dataset instance.
|
Type
|
Dataset
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
FileDataset
Default constructor. Notice that the following symbols are used by default:
• sCommentValue: "%"
• sSeparationSymbol: ","
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
FileDataset
This constructor receives the name of the file as parameter. The following symbols are used as default:
• sCommentValue: "%"
• sSeparationSymbol: ","
Parameter:
• sFileName: The filename of
the dataset source.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFileName : String
|
getCommentValue
Getter method of
the property sCommentValue.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getFileName
Getter method of
the filename of
the dataset source.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getSeparationSymbol
Getter method of
the property sSeparationSymbol.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
readDataset
Implementations of
this abstract method will read the dataset from the file specified by the constructor.
Parameters:
• sContentFormat: String that specifies the reading format of
the dataset file. Construct the
string using a sequence of control tokens:
o % to omit a line (only one line).
o %name to read the name of columns (only one line).
o %col to read data (zero, one or more lines).
Example: the string “%%%col%%name”
indicates that the first two lines must be omitted, then data is
read and, finally, the last line will contain the column names.
• sColumnFormat: A String that contains an ordered sequence of tokens that determine the data type of each column to be read. Use the following tokens:
o s: Nominal column
o f: Real column
o c: Categorical column
o b: Binary column
o i: Integer column
o %: Skip this column (the column skipped is
not
processed)
Additionally, notice that other tokens can be considered depending of the
specific dataset
source (e.g., d for columns of
type date).
Exceptions:
• FileNotFoundException
• IOException
• IllegalFormatSpecificationException
• NotAddedValueException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• inout sColumnFormat : String
• inout sContentFormat : String
|
setCommentValue
Setter method of the
property sCommentValue.
Parameter:
• sComment: The token/string indicating the symbol that represents a comment line in the dataset
file.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sComment : String
|
setFileName
Setter method of the
property sFileName. Parameter:
• sFileName: The filename of
the dataset source.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFileName : String
|
setSeparationSymbol
Setter method of the
property sSeparationSymbol. Parameter:
• sSeparationSymbol: The token used to differentiate between instances in the same line of
the dataset source.
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sSeparator : String
|
writeDataset
This abstract method defines the signature of
the write method for every file dataset. Implementations of
this method deal with the serialization (writing) of the current column structure into each specific file format.
Parameter:
• sOutputFile: The path where the dataset file will be
saved.
Exception:
• IOException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• inout sOutputFile : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• CsvDataset
|
|
Name
|
|
|
Related Element
|
• ExcelDataset
|
|
Name
|
|
|
Related Element
|
• ArffDataset
|
|
Name
|
|
|
Related Element
|
• Dataset
|
Class
InstanceIterator
InstanceIterator is the class that implements the interface IIterator for covering the instances of
the dataset. Thus, this class represents an
iterator to access each row/instance in
a dataset. The instance
iterator provides methods to cover the whole set of instances and keeps the reference to the
dataset being
iterated.
Figure
18. Class InstanceIterator
|
Name
|
InstanceIterator
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::InstanceIterator
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
|
|
Realized Interface
|
• IIterator
|
Attribute Detail
All attributes are private.
Operation Detail
currentInstance
This method returns the list of objects that form the currently pointed instance in the dataset.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
first
This method returns the list of objects that form the first instance in the dataset and sets the pointer to the first instance.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
InstanceIterator
Default iterator constructor.
Parameter:
• oDataset: The dataset to be covered by the iterator.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDataset : Dataset
|
isDone
This method returns true if the dataset has no more instances to be iterated. False, otherwise.
|
Type
|
boolean
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
next
This method increases the instance pointer by one, i.e. sets the pointer to the next instance in the
dataset.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Interface Realization
|
Name
|
|
|
Related Element
|
• IIterator
|
Interface
IIterator
IIterator is the interface that any instance iterator has to implement, as InstanceIterator does.
Figure
19. Interface IIterator
|
Name
|
IIterator
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::IIterator
|
|
Visibility
|
public
|
|
Base Classifier
|
|
Operation Detail
currentInstance
The implementation of this method has to return the current pointed instance in the dataset as a List of instances of any class from Object.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
first
An implementation of this
method returns the first instance of the
dataset. From here on, the current
instance pointed by the iterator should be
this first one.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
isDone
This method should be implemented to return True if the iterator points to the
last instance of the dataset. It returns False otherwise.
|
Type
|
boolean
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
next
The implementation of this method increases the iterator to the next instance in the dataset.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
Relation Detail
Interface Realization
|
Name
|
|
|
Related Element
|
• InstanceIterator
|
This package contains the classes related to the different types of columns supported by the library. At the
moment, datapro4j provides an implementation for the following types:
• Binary column, for positive or negative values.
• Categorical column, for prefixed string values, considered as an enumeration of
categories.
• Date column.
• Integer column, for numerical integer values.
• Nominal column, for free valued strings.
• Numerical column, for numerical real values.
• Range column, for those values that represent a numerical interval (minimum, maximum), where
both open and close ranges can be considered.
Columns are coded following the philosophy of
the bridge design pattern, where an abstraction is decoupled
from its implementation. In
this way, the programmer can add to the library new implementations of some of
the columns provided, e.g. for performance reasons, without altering the manner in which the rest of the library–including algorithms–interacts with this column.
Therefore, every column type is implemented by at
least two different classes: its abstraction, where the
accessor methods to its functionalities exist, and its implementation, where these functionalities are coded,
and invoked from the abstraction.
Using columns properly demands considering the following rules:
• Any code from
the library (i.e. from other
columns, datasets or strategies) should always invoke
methods of the abstraction. Never invoke directly to the column implementation (only its own
abstraction should).
• Altering current abstractions may cause unexpected failures. Use generalization or provide conversion methods to build your own abstractions instead.
• Abstractions and implementations must be subclasses of ColumnAbstraction and ColumnImplementation, respectively.
• Datapro4j only supports one implementation class per abstraction. If the programmer wants to
have
more than one implementation, then more than one abstraction should be provided, or a factory pattern should be coded.
• If new abstractions (i.e. type of
columns) are provided, modify the enumeration ColumnType
accordingly.
Figure
20. Package es.uco.kdis.datapro.dataset.Column
|
Name
|
Column
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column
|
Class
ColumnAbstraction
This abstract class implements the common functionalities contained by every column in the dataset. It also
defines the methods that are not coded by the implementation class, but they refer to the column metainformation (e.g. name, type, etc.). The latter methods are directly implemented by abstractions, since
they
do not require any access to data.
Figure
21. Abstract class ColumnAbstraction
|
Name
|
ColumnAbstraction
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::ColumnAbstraction
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Attribute Detail
ctColumnType
The column type, as represented by the enumeration defined by the class ColumnType.
|
Type
|
ColumnType
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
1
|
oImpl
A reference to the
implementation object.
|
Type
|
ColumnImpl
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
1
|
sName
The name of
the column.
|
Type
|
String
|
|
Default Value
|
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
addAllValues
This method calls the implementation to add a list of
values at the end of the column.
Parameter:
• rgoCol The list of values to be
added. The objects here contained must satisfy the type required by the column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoCol : List<Object>
|
addValue
This method calls the implementation to add a single value at the
end of the column.
Parameter:
• oValue The value to be added. It must satisfy the type required by the column.
The method returns the number of items successfully added to
the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
This method calls the implementation to add a single value at the
end of the column.
Parameters:
• oValue The value to be added. It must satisfy the type required by the column.
• bForce is used to indicate that the value must be added, independently of
the constraints and
addition policies defined by the column type.
The method returns the number of items successfully added to
the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bForce : boolean
• inout oValue : Object
|
addValue
This method calls the implementation to add a single value at a given position in the column.
Parameters:
• iIndex indicates the element position where the item has to be added.
• oValue The value to be added. It must satisfy the type required by the column.
The method returns the number of items successfully added to
the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
ColumnAbstraction
Default constructor with parameters. Subclasses may override this method or create new constructors.
This constructor only assigns the parameter values to
its
respective variables. The constructor in the subclass should create the implementation object and assigned it to the variable oImpl.
Parameters:
• ctColumnType The column type.
• sName The Name of the column to be created.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout ctColumnType : ColumnType
• inout sName : String
|
countEmptyValues
This method calls the implementation to return the number of empty values in the column set.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
This method calls the implementation to return the number of invalid values (i.e. empty, null and missing
values) in the column set.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
This method calls the implementation to return the number of missing values in the column set.
|
Type
|
Int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
This method calls the implementation to return the number of null values in
the
column set.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
This method calls the implementation to return the element at the given position.
Parameter:
• iPos Position of the
element queried.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getEmptyValue
This method calls the implementation to return the column-specific empty value. This is
not
the default
empty value used by datapro4j
(Class EmptyValue) for reading, writing or internally checking empty
values, but it serves the developer to define its own use (e.g., the symbol associated to
the empty value, or whatever).
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMissingValue
This method calls the implementation to return the column-specific missing value. This is not the default missing value used by datapro4j (Class MissingValue) for reading, writing or internally checking
missing values, but it serves the developer to define its own use (e.g., the symbol associated to a missing value, or whatever).
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getName
This method returns the name given of the
column.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getNullValue
This method calls the implementation to return the column-specific null value. This is
not the default null value used by datapro4j (Class NullValue) for reading, writing or internally checking null values, but it serves the developer to
define its own null object.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getSize
This method calls the implementation to return the size of
the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getType
This method returns the type of
the column as a value of
ColumnType.
|
Type
|
ColumnType
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
This method calls the implementation to return the list of items (as instances of
Object) contained in
the column.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
removeValue
It calls the implementation to remove an
element in the column at a given position.
Parameter:
• iIndex The index of the element to be removed.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
setEmptyValue
This method calls the implementation to set the column-specific empty value, if required. This is not the
default empty value used by datapro4j (Class EmptyValue) for reading, writing or internally checking empty values, but the developer has to define its usage in
the code of the proper strategies.
Parameter:
• oEmptyValue The empty value to be set.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oEmptyValue : Object
|
setMissingValue
This method calls the implementation to set the column-specific missing value, if required. This is not the default missing value used by datapro4j (Class MissingValue) for reading, writing or internally
checking missing values, but the developer has to define its usage in
the code of the proper strategies. Parameter:
• oMissingValue The missing value to
be set.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oMissingValue : Object
|
setName
This method sets the name of the column.
Parameter:
• sName The new name for the column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
setNullValue
This method calls the implementation to set the column-specific null value, if required. This is not the default null value used by datapro4j (Class NullValue) for reading, writing or internally checking null
values, but the developer has to define its usage in the code of the proper strategies.
Parameter:
• oNullValue The null value to
be set.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oNullValue : Object
|
setValue
This method calls the implementation to set the value of an
element in the column at a given position.
Parameters:
• oValue The value to be added.
• iIndex The element position in the column.
It returns the number of
elements correctly added.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
Relation Detail
Association
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
|
Name
|
|
|
Related Element
|
• ColumnType
|
|
Name
|
rgoColumns
|
|
Related Element
|
• Dataset
|
Generalization
|
Name
|
|
|
Related Element
|
• CategoricalColumn
|
|
Name
|
|
|
Related Element
|
• NumericalColumn
|
|
Name
|
|
|
Related Element
|
• DateColumn
|
|
Name
|
|
|
Related Element
|
• BinaryColumn
|
|
Name
|
|
|
Related Element
|
• NominalColumn
|
|
Name
|
|
|
Related Element
|
• RangeColumn
|
Class
ColumnImpl
This abstract class serves as a base for column implementation classes. These classes comprise the real code accessing data in the column. Only metainformation is
managed by its abstraction.
Note: None of
its
methods should be
directly invoked, apart from its specific abstraction. Thus, for a given column type, abstraction is inalterable, whereas implementation could be adapted by the programmer.
Figure
22. Abstract class ColumnImpl
|
Name
|
ColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::ColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Attribute Detail
oEmptyValue
This object represents a column-specific empty value. Notice that this is
not
the standard empty value
object, as used by datapro4j strategies and datasets.
|
Type
|
Object
|
|
Default Value
|
null
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oMissingValue
This object represents a column-specific missing value. Notice that this is not the standard missing value
object, as used by datapro4j strategies and datasets.
|
Type
|
Object
|
|
Default Value
|
null
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
oNullValue
This object represents a column-specific null value. Notice that this is
not
the standard null value object,
as used by datapro4j
strategies and datasets.
|
Type
|
Object
|
|
Default Value
|
null
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
The following methods
code the implementation for their corresponding abstraction methods.
addAllValues
This method implements the method addAllValues of the column abstraction, returning the number of
objects successfully added.
Parameter:
• rgoCol The list of item objects to be added to the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• inout rgoCol : List<Object>
|
addValue
This method implements the method addValue of the column abstraction, returning the number of objects successfully added.
Parameter:
• oValue The value to be added.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• inout oValue : Object
|
addValue
This method implements the method addValue of the column abstraction, returning the number of objects successfully added.
Parameters:
• oValue The value to be added
• bForce If true, the implementation must force its addition.
Note: By default bForce is not considered. Otherwise, the subclass implementing the specific
column should explicitly rewrite this method.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
• in bForce : boolean
|
addValue
This method implements the method addValue of the column abstraction, returning the number of objects successfully added.
Parameters:
• oValue The value to be added.
• iIndex The position in
the column to add the value.
|
Type
|
Int
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• inout oValue : Object
• in iIndex : int
|
countEmptyValues
This method implements the method countEmptyValue of the column abstraction, returning the number of empty values contained in
the column values. -1 is returned if this value could not be
calculated.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
This method implements the method countInvalidValue of the column abstraction, returning the number of invalid values (null, empty and missing values) contained in
the column values. -1
is returned
if
this value could not be calculated.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
This method implements the method countMissingValue of the column abstraction, returning the number of missing values contained in the column values. -1 is
returned if this value could not be
calculated.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
This method implements the method countNullValue of the column abstraction, returning the number
of null values contained in the column values. -1 is
returned if this value could not be calculated.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
This method implements the method getElement of the column abstraction, returning the element at
the given position.
Parameter:
• iPos The position of the
element to be returned.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• in iPos : int
|
getEmptyValue
This method implements the method getEmptyValue of the column abstraction, returning the element representing the column-specific empty value.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMissingValue
This method implements the method getMissingValue of the column abstraction, returning the element representing the column-specific missing value.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getNullValue
This method implements the method getNullValue of the column abstraction, returning the element
representing the column-specific null value.
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getSize
This method implements the method getSize of the column abstraction, returning the number of
elements contained in the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
getValues
This method implements the method getValues of the column abstraction, returning the list of elements (as instances of Object) contained in the
column.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
|
removeValue
This method implements the method removeValue of the column abstraction.
Parameter:
• iIndex The position in
the column to add the value.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• in iIndex : int
|
setEmptyValue
This method implements the method setEmptyValue of the column abstraction, setting the element representing the column-specific empty value.
Parameter:
• oEmptyValue The object representing a specific empty value in this column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oEmptyValue : Object
|
setMissingValue
This method implements the method setMissingValue of the column abstraction, setting the element representing the column-specific missing value.
Parameter:
• oMissingValue The object representing a specific missing value in this column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oMissingValue : Object
|
setNullValue
This method implements the method setNullValue of the column abstraction, setting the element
representing the column-specific null value.
Parameter:
• oNullValue The object representing a specific null value in this column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oNullValue : Object
|
setValue
This method implements the method setValue of the column abstraction, setting the element value at
the given position.
Parameters:
• oValue The object value to set.
• iIndex The position index in the column.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
true
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
Relation Detail
Association
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Generalization
|
Name
|
|
|
Related Element
|
• RangeColumnImpl
|
|
Name
|
|
|
Related Element
|
• NominalColumnImpl
|
|
Name
|
|
|
Related Element
|
• NumericalColumnImpl
|
|
Name
|
|
|
Related Element
|
• DateColumnImpl
|
|
Name
|
|
|
Related Element
|
• CategoricalColumnImpl
|
|
Name
|
|
|
Related Element
|
• BinaryColumnImpl
|
Enumeration ColumnType
This enumeration contains the different types of columns supported by datapro4j. The following types are
currently supported:
• Binary
• Categorical
• Date
• Integer
• Nominal
• Numerical
• Range
Note: If the programmer wants to check the column type, the following code should be used (e.g. for binary
columns)
ColumnAbstraction oCol;
…
if (oCol.getType().equals(ColumnType.Binary)) {
…
}
Figure
23. Enumeration ColumnType
|
Name
|
ColumnType
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::ColumnType
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Attribute Detail
Binary
Boolean attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Categorical
Categorical attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Date
Date attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Integer
Integer attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Nominal
Nominal attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Numerical
Numerical attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Range
Range attribute
|
Type
|
|
|
Default Value
|
|
|
Visibility
|
public
|
|
Multiplicity
|
|
Relation Detail
Association
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
BinaryColumn
This class represents the abstraction of
a binary column. Here the methods that provide specific operations
on specific binary data are defined.
Figure
24. Class BinaryColumn
|
Name
|
BinaryColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::BinaryColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnAbstraction
|
|
Realized Interface
|
|
Operation Detail
BinaryColumn
Default constructor. The implementation BinaryColumnImpl is invoked.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
BinaryColumn
Constructor with the name of the column as a parameter. The implementation BinaryColumnImpl is invoked.
Parameter:
• sName The name of the column.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
toCategorical
This method calls the implementation to return a categorical column generated from the binary column.
The
resulting categorical column defines two categories, one per each binary value (false, true).
Parameters:
• sFalseCategory The category representing the false binary value.
• sTrueCategory The category representing the true binary value.
Notes:
• If the value is an empty or a missing value, then a false value is considered.
• If the value is a null value, then a null value is considered.
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFalseCategory : String
• inout sTrueCategory : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
BinaryColumnImpl
This class provides the implementation code accessing real data in a binary column. Binary values are stored as objects of class Boolean.
Note: None of
its
methods should be
directly invoked, but only from its specific abstraction.
Figure
25. Class BinaryColumnImpl
|
Name
|
BinaryColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::BinaryColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnImpl
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
For a more complete specification of the methods inherited from ColumnImpl, see its specifications above.
addAllValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoCol : List<Object>
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
BinaryColumnImpl
Default constructor.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countEmptyValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getSize
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
removeValue
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iIndex : int
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
toCategorical
This method implements the method toCategorical of the binary column abstraction, converting the
binary column into a categorical column.
Parameters:
• sName The name of
the column. By default this property is set by the abstraction to the current
name
of the binary column.
• sFalseCategory The category representing the false binary value.
• sTrueCategory The category representing the true binary value.
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
• inout sFalseCategory : String
• inout sTrueCategory : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
Class
CategoricalColumn
This class defines the abstraction of
a categorical column, where every value belongs to a predefined category. Here the methods that provide specific operations on categorical data are defined.
Figure
26. Class CategoricalColumn
|
Name
|
CategoricalColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::CategoricalColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnAbstraction
|
|
Realized Interface
|
|
Operation Detail
addCategory
This method calls the implementation to add a new category to the set of allowable values. Categories
are
included as objects of
class String.
Parameter:
• szCategory The new category in the
column
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout szCategory : String
|
CategoricalColumn
Constructor with the name of the column as a parameter. The implementation
CategoricalColumnImpl is invoked.
Parameter:
• sName The name of the column
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
CategoricalColumn
Default constructor. The implementation CategoricalColumnImpl is invoked.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getCategoryIndex
This method calls the implementation to return the index in the list of categories of
a given string. The value -1
is returned if the value is not found.
Parameter:
• szCategory The string representing the category to
be searched in the list of categories
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout szCategory : String
|
getCategoryList
This method calls the implementation to return the list of categories in the column.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getCategoryName
This method calls the implementation to return the category string stored in a given position of
the list of
categories. null is returned if the index given is not
valid.
Parameter:
• iIndex The index of the
wanted category
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iIndex : Integer
|
getElementIndex
This method calls the implementation to return the element stored in
a given position in
the column. The category index is returned, whereas the default method getElement (inherited from ColumnAbstraction) returns the category by name. If the value is invalid, -1 is returned.
Parameter:
• iPos The index of the
item in the column
Exceptions:
• IndexOutOfBoundsException
|
Type
|
Integer
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
replaceCategory
This method calls the implementation to replace a given category with a new one. Parameters:
• szOldCategory The category string to be replaced
• szNewCategory The new category string to
be set
• bJoinCategory If the new category string already exists, then this
parameter determines whether the values in of the
old category are mixed together with the values of
the column whose values coincide
1 is returned if the category is
successfully replaced, or 0 otherwise.
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bJoinCategory : boolean
• inout szNewCategory : String
• inout szOldCategory : String
|
toBinary
This method calls the implementation to return a binary column generated from the categorical column. Invalid values remain unaltered.
Parameter:
• aReferenceTrueValues The list of category strings to be
as true values
|
Type
|
BinaryColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout aReferenceTrueValues : List<String>
|
toNominal
This method calls the implementation to return a nominal column generated from the strings stored in
the categorical column. Nominal values are extracted from
the strings representing each category.
|
Type
|
NominalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toNumerical
This method calls the implementation to return an integer column generated from
the index values
assigned to
the categories in the source column.
|
Type
|
IntegerColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
CategoricalColumnImpl
This class provides the implementation code accessing real data in a categorical column. Categories are stored as a HashMap between a String and an Integer. Thus, internally, data are stored as an ArrayList of Integer, whereas their correspondences to categories are saved as String.
This class should never be
directly invoked, apart from those invocations coming from
its
abstraction.
Figure
27. Class CategoricalColumnImpl
|
Name
|
CategoricalColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::CategoricalColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnImpl
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
For a more complete specification of the methods inherited from ColumnImpl, see its specification above.
Notice that values can be added both as a
String –identifier- and as an Integer–index- (see methods
addValue, addAllValues). In both cases only elements belonging to
valid categories are added to the set of values in the column.
addCategory
This method implements the functionality of addCategory in the categorical column abstraction, adding
a new category to the column. This category should not exist. It returns the index of the new category, if
successfully created, or -1
if
the category cannot be added.
Parameter:
• sCat The identifier of
the
new category
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sCat : String
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bForce : boolean
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
CategoricalColumnImpl
Default constructor.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countEmptyValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getCategoryIndex
This method implements the functionality of getCategoryIndex in the column abstraction, returning
the index of the category passed as String, or -1 if the category does not exist in the list of categories of
the column.
Parameter:
• sCategory The category identifier
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sCategory : String
|
getCategoryList
This method implements the functionality of getCategoryIndex in the column abstraction, returning
the list of category identifiers comprised by the category list. The resulting list is not sorted.
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getCategoryName
This method implements the functionality of getCategoryName in the column abstraction, returning the
identifier of
the category whose index is passed as parameter. If the category does not exist, then null is returned.
Parameter:
• iIndex The category index
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iIndex : Integer
|
getElement
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getElementIndex
This method implements the functionality of getElementIndex in the column abstraction, returning the
category index stored at
a given position. Notice that indexes in the category list
do not have to
be sorted or sequencial, since categories may be successively created and deleted, causing gaps in
the index sequence. Always consider category indexes as numerical identifiers, never as sequential indexes.
This method returns -1 if the position given is
invalid.
Parameter:
• iPos The position given in
the category list.
Exceptions:
• IndexOutOfBoundsException
|
Type
|
Integer
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getSize
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
removeValue
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
replaceCategory
This method implements the functionality of replaceCategory in the column abstraction, updating
both the category list
and replacing the values in the column. 1 is returned if done; 0, otherwise.
Parameters:
• sOldCategory The old category identifier to
be replaced
• sNewCategory The new category
• bJoinCategory If true, if the new category identifier already exists in the column, then the values with the old category identifier will be joined to the already existing identifier, having only one category as a result
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bJoinCategory : boolean
• inout sNewCategory : String
• inout sOldCategory : String
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
toBinary
This method implements the functionality of toBinary in the column abstraction, returning a binary
column constructed from
the data contained in the categorical column. The list of category identifiers
considered as True values in the binary column is passes as parameter. The non included category identifiers are considered as False values. Note that invalid values are observed.
Parameters:
• aReferenceTrueValues The list of categories representing true values
• sName The name of the new binary column
|
Type
|
BinaryColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout aReferenceTrueValues : List<String>
• inout sName : String
|
toNominal
This method implements the functionality of toNominal in the column abstraction, returning a nominal column constructed from the data contained in the categorical column. Strings for the nominal column are constructed from the category identifiers.
Parameter:
• sName The name of the new nominal column
|
Type
|
NominalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
toNumerical
This method implements the functionality of toNumerical in the column abstraction, returning an
integer column constructed from
the data contained in
the categorical column. Numbers of the integer
column are extracted from
the
category indexes.
Parameter:
• sName The name of the new integer column
|
Type
|
IntegerColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• RangeColumnImpl
|
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
Class
DateColumn
This class represents the abstraction of
a date datatype column. This type of column is specifically required by ARFF datasets.
Figure
28. Class DateColumn
|
Name
|
DateColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::DateColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnAbstraction
|
|
Realized Interface
|
|
Operation Detail
addDateSpecification
This method calls the implementation to set the date format specification of
the values in
the column.
Parameter:
• sDate The format specification of
the values in the
date column
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDate : SimpleDateFormat
|
DateColumn
Default constructor with no parameters. The implementation DateColumnImpl is invoked.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
DateColumn
Constructor with the name of the column as a parameter. The implementation DateColumnImpl is invoked.
Parameter:
• sName The name of the column
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
getDateSpecification
This method calls the implementation to get the date format specification of
the values in
the column.
|
Type
|
SimpleDateFormat
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
DateColumnImpl
This class provides the implementation code accessing real data in a date column. Values are stored as
Date objects according to
the
format specified by a given SimpleDateFormat object. This class should not be invoked directly, only by the column abstraction.
Figure
29. Class DateColumnImpl
|
Name
|
DateColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::DateColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnImpl
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
For a more complete specification of the methods inherited from ColumnImpl, see its specifications above.
addAllValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoCol : List<Object>
|
addDateSpecification
This method implements the method addDateSpecification of
the date column abstraction, setting the date format specification of
the values in the column.
Parameter:
• sDate The format specification of
the values in the
date column
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oDate : SimpleDateFormat
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bForce : boolean
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
DateColumnImpl
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getDateSpecificaiton
This method implements the method getDateSpecification of the column abstraction, returning the date
format specification of
the
values in
the column.
|
Type
|
SimpleDateFormat
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in
iPos : int
|
getSize
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
removeValue
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
Class
IntegerColumn
This class represents the abstraction of
an integer column. Integer columns are a specialization of numerical (real) columns.
Figure
30. Class IntegerColumn
|
Name
|
IntegerColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::IntegerColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• NumericalColumn
|
|
Realized Interface
|
|
Operation Detail
Many methods are specializations of
their respective methods
in the numerical column
(NumericalColumn), adapted to the domain of
integer values.
getiMaxInterval
Analogously to getdMaxInterval in the NumericalColumn abstraction class, this method gets the maximum integer
value allowed for this column.
|
Type
|
Integer
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getiMinInterval
Analogously to getdMinInterval in the NumericalColumn abstraction class, this method gets the minimum integer value allowed for this column.
|
Type
|
Integer
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMaxValue
See getMaxValue in the specification of the NumericalColumn abstraction class.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMinValue
For further information, see getMinValue in the specification of
the NumericalColumn abstraction class.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
IntegerColumn
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
IntegerColumn
Constructor with the name of the resulting column as a parameter.
Parameter:
• sName The Name of the column
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
mean
For further information, see mean in the specification of
the NumericalColumn abstraction class.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setiMaxInterval
Analogously to setdMaxInterval in the NumericalColumn abstraction class, this method sets the
maximum integer value allowed for this column.
Parameter:
• iMaxInterval The maximum value allowed in the
column
Exceptions:
• IllegalAccessException if the value cannot be
set.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iMaxInterval : Integer
|
setiMinInterval
Analogously to setdMinInterval in the NumericalColumn abstraction class, this method sets the
minimum integer
value allowed for this column.
Parameter:
• iMinInterval The maximum value allowed in the
column
Exceptions:
• IllegalAccessException if the value cannot be
set.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iMinInterval : Integer
|
standardDeviation
For further information, see standardDeviation in the specification of
the NumericalColumn abstraction class.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toCategorical
This method calls the implementation to return a categorical column using the values contained in the integer column, where each different value constitutes a different category.
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toNumerical
This method calls the implementation to return a numerical column using the values contained in the
integer column, where each integer value is casted to a double value.
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• NumericalColumn
|
Class
IntegerColumnImpl
This class provides the implementation code accessing real data in an integer column. This class is
a specialization of the numerical column implementation (NumericalColumnImpl). Integer values are stored
as objects of class Integer. This class and its methods should not be invoked directly.
Figure
31. Class IntegerColumnImpl
|
Name
|
IntegerColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::IntegerColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• NumericalColumnImpl
|
|
Realized Interface
|
|
Operation Detail
For further information, see a complete specification of these methods in NumericalColumnImpl and ColumnImpl.
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
getMaxValue
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMinValue
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
IntegerColumnImpl
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
mean
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
standardDeviation
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toCategorical
This method implements the method toNumerical of the abstraction, returning a categorical column using the values contained in the integer column, where each different value constitutes a different category.
Parameter:
• sName The name of the resulting column
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
toNumerical
This method implements the method toNumerical of the abstraction, returning a numerical column using
the values contained in the integer column, where each different value constitutes a different category.
Parameter:
• sName The name of the resulting column
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• NumericalColumnImpl
|
Class
NominalColumn
This class represents the abstraction of
a nominal column containing free-style strings as values. Here the methods that provide specific operations of
nominal values are defined.
Figure
32. Class NominalColumn
|
Name
|
NominalColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::NominalColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnAbstraction
|
|
Realized Interface
|
|
Operation Detail
NominalColumn
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
NominalColumn
Constructor with the name of the column as parameter.
Parameter:
• sName Name of the column
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
toCategorical
This method calls the implementation to return a categorical column, where each different string is a category (no repetition).
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toNumerical
This method calls the implementation to return a numerical column, where each string is
parsed to a numerical value. If the string could not be parsed, then an empty value is
added instead. Notice that
upper and lower interval values are not set for the numerical column returned.
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
NominalColumnImpl
This class provides the implementation code accessing real data in the nominal column. Nominal values are stored as String objects. Note that these methods should not be invoked directly.
Figure
33. Class NominalColumnImpl
|
Name
|
NominalColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::NominalColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnImpl
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
For a more detailed specification
of the methods inherited from ColumnImpl, see its
specification above.
addAllValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoCol : List<Object>
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
countEmptyValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getSize
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
NominalColumnImpl
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
removeValue
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
toCategorical
This method implements the method toCategorical of the abstraction, returning a categorical column, where each different string is
a category (no repetition).
Parameter:
• sName The name of the column to be created
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
toNumerical
This method implements the method toNumerical of the abstraction, returning a numerical column,
where each string is parsed to a numerical value. If the string could not be parsed, then an
empty value is
added
instead. Notice that upper and lower interval values are not set for the numerical column returned.
Parameter:
• sName The name of the column
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
Class
NumericalColumn
This class represents the abstraction of
a numerical (real) column.
Figure
34. Class NumericalColumn
|
Name
|
NumericalColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::NumericalColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnAbstraction
|
|
Realized Interface
|
|
Attribute Detail
dMaxInterval
This attribute indicates the maximum value allowed in
the column. This property should be accessed using getter/setter methods.
|
Type
|
Double
|
|
Default Value
|
Double.MAX_VALUE
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
dMinInterval
This attribute indicates the minimum value allowed in the column. This property should be accessed
using getter/setter methods.
|
Type
|
Double
|
|
Default Value
|
Double.MIN_VALUE
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
getdMaxInterval
This method returns the maximum real value allowed in the column. The implementation is not invoked,
since this information is part of metadata.
|
Type
|
Double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getdMinInterval
This method returns the minimum real value allowed in the column. The implementation is
not
invoked, since this information is part of metadata.
|
Type
|
Double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMaxValue
This method calls the implementation to get the maximum existing value in the column data.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMinValue
This method calls the implementation to get the minimum existing value in the column data.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
mean
This method calls the implementation to get the mean value of
the column data.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
normalize
This method calls the implementation to normalize the set of
values in the numerical column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
NumericalColumn
Default constructor with no parameters. The implementation NumericalColumnImpl is invoked.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
NumericalColumn
Constructor with the name of the column as a parameter. The implementation NumericalColumnImpl is invoked.
Parameter:
• sName The name of the column
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
setdMaxInterval
This method sets the maximum real value allowed in the column. The implementation is not invoked, since this information is part of
the column metadata.
Parameter
• dMaxInterval The maximum value allowed
Exceptions:
• IllegalAccessException if the value cannot be
set
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout dMaxInterval : Double
|
setdMinInterval
This method sets the minimum real value allowed in the column. The implementation
is not invoked,
since this information is part of
the column metadata.
Parameter
• dMinInterval The minimum value allowed
Exceptions:
• IllegalAccessException if the value cannot be
set
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout dMinInterval : Double
|
standardDeviation
This method calls the implementation to return the standard deviation calculated from
the set of values in the numerical column.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
standarize
This method calls the implementation to standarize the set of values in the numerical column.
Parameters:
• dMean Value of
the mean used to standardize the set of values of
the column
• dVariance Value of
the variance used for the standardization
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in
dMean : double
• in
dVariance : double
|
toInteger
This method calls the implementation to return an integer column containing values extracted from the numerical column. It returns an IntegerColumn object.
Parameter:
• bRoundedValue if false, values are truncated; if true,
values are rounded.
|
Type
|
IntegerColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bRoundedValue : boolean
|
toNominal
This method calls the implementation to return a nominal column, where strings are constructed from
real
values.
|
Type
|
NominalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
NumericalColumnImpl
This class provides the implementation code accessing real data in a numerical column. Values are stored
as objects of the
class Double. Notice that this class should not be directly instantiated, with the exception
of its abstraction.
Figure
35. Class NumericalColumnImpl
|
Name
|
NumericalColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::NumericalColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnImpl
|
|
Realized Interface
|
|
Attribute Detail
All the attributes are either private or protected.
Operation Detail
addAllValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoCol : List<Object>
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
countEmptyValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getMaxValue
This method implements the method getMaxValue of the abstraction class, returning the maximum
existing value in the column.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMinValue
This method implements the method getMinValue of the abstraction class, returning the maximum
existing value in the column.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getSize
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
mean
This method implements the method mean of the abstraction class, returning the mean value of the
column.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
normalize
This method implements the method normalize of the abstraction class, calculating and normalizing
the values contained in the set of
values of
the
column.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
NumericalColumnImpl
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
removeValue
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
standardDeviation
This method implements the method standardDeviation of the abstraction class, returning the
standard deviation value of the
set of values contained in the numerical column.
|
Type
|
double
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
standarize
This method implements the method standarize of the abstraction class, standardizing the values in
the column according to the mean and variance passed as parameter.
Parameters:
• dMean Mean value considered for the standardization
• dVariance Variance value considered for the standardization
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dMean : double
• in dVariance : double
|
toInteger
This method implements the method toInteger of the abstraction class, returning an integer column
calculated from
the numerical column.
Parameters:
• sName The name of the resulting new column
• bRoundedValue If false, values are truncated; if true,
values are rounded
|
Type
|
IntegerColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in bRoundedValue : boolean
• inout sName : String
|
toNominal
This method implements the method toNominal of the abstraction class, returning a nominal column
which strings are constructed parsing the numerical values in the column.
Parameter:
• sName The name of the resulting new column
|
Type
|
NominalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
Class
RangeColumn
This class represents the abstraction of
a range column, whose values are intervals with a minimum and a
maximum value in the range.
Figure
36. Class RangeColumn
|
Name
|
RangeColumn
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::RangeColumn
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnAbstraction
|
|
Realized Interface
|
|
Operation Detail
RangeColumn
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
RangeColumn
Constructor with the name of the column as a parameter.
Parameter:
• sName The name of the column.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
toCategorical
This method calls the implementation to return a categorical column extracted from
the range data
contained in the
column. The method returns a CategoricalColumn object.
Exceptions:
• NotAddedValueException
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toNumerical
This method calls the implementation to return a numerical column extracted from the range values contained in the
column, and according to on
of the following modes:
0: The minimum value of
each range is selected.
1: The maximum value of
each range is selected.
2: The mean value between min and max is selected.
3: A random value in the
range is selected.
It returns the resulting NumericalColumn object.
Parameter:
• iMode An integer between 0 and 3 indicating the conversion mode, as described above.
Exceptions:
• NotAddedValueException
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout iMode : int
|
toNumericalByGaussian
This method calls the implementation to return a numerical column extracted from the range values contained in
the column, according to the Gauss distribution.
Parameters:
• dMean The arithmetic mean for the distribution
• dStdDev The standard deviation for the distribution
It returns the resulting NumericalColumn object.
Exceptions:
• NotAddedValueException
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dMean : double
• in dStdDev : double
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnAbstraction
|
Class
RangeColumnImpl
This class, the abstraction of
a range column (i.e. a representation of
a [min, max] interval),
is the one that should be
used
by the programmer, since it hides the actual implementation
of the column. Even when the implementation changes, the abstraction must remain unaltered.
Figure
37. Class RangeColumnImpl
|
Name
|
RangeColumnImpl
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Column::RangeColumnImpl
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ColumnImpl
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
For a detailed specification of the
methods inherited from ColumnImpl, see its specifications above.
addAllValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout rgoValues : List<Object>
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oValue : Object
|
addValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
countEmptyValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countInvalidValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countMissingValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
countNullValues
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getElement
|
Type
|
Object
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iPos : int
|
getSize
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getValues
|
Type
|
List<Object>
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
RangeColumn
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
RangeColumn
Constructor with the name of the column as a Parameter.
Parameter:
• sName The name of the column.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sName : String
|
removeValue
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
|
setValue
|
Type
|
int
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iIndex : int
• inout oValue : Object
|
toCategorical
This method implements the method toCategorical of the abstraction, returning a categorical column extracted from
the range data contained in
the column. The method returns the resulting CategoricalColumn object.
Exceptions:
• NotAddedValueException
|
Type
|
CategoricalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
toNumerical
This method implements the method toNumerical of the abstraction, returning a numerical column extracted from the range values contained in the column, and according to on of the following modes:
0: The minimum value of
each range is selected.
1: The maximum value of
each range is selected.
2: The mean value between min and max is selected.
3: A random value in the
range is selected.
It returns the resulting NumericalColumn object.
Parameter:
• iMode An integer between 0 and 3 indicating the conversion mode, as described above.
Exceptions:
• NotAddedValueException
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in iMode : int
|
toNumericalByGaussian
This method implements the method toNumericalByGaussian of the abstraction, returning a numerical column extracted from the range values contained in the column, according to the Gauss distribution.
Parameters:
• dMean The arithmetic mean for the distribution
• dStdDev The standard deviation for the distribution
It returns the resulting NumericalColumn object.
Exceptions:
• NotAddedValueException
|
Type
|
NumericalColumn
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dMean : double
• in dStdDev : double
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ColumnImpl
|
Figure
38. Package es.uco.kdis.datapro.dataset.Source
|
Name
|
Source
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Source
|
Class
ArffDataset
ArffDataset implements the ARFF (Attribute-Relation File Format) dataset file specification, as used by Weka. This is a subclass of
FileDataset.
ARFF files are ASCII text files that describe a list of instances sharing a set of attributes. After a few heading
lines, where the metainformation is
presented, one instance per line is
dumped, until the end of the
file is reached.
Types of attribute in
ARFF dataset files:
• @ATTRIBUTE name numeric (As numerical columns)
• @ATTRIBUTE name {value1,
value2, ...} (As categorical columns)
• @ATTRIBUTE name string (As nominal columns)
• @ATTRIBUTE name date "yyyy-MM-dd HH:mm:ss" (As date columns)
For a further description, visit the web site http://www.cs.waikato.ac.nz/ml/weka/arff.html (Nov. 1st, 2008).
Figure
39. Class ArffDataset
|
Name
|
ArffDataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Source::ArffDataset
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• FileDataset
|
|
Realized Interface
|
|
Attribute Detail
Some attributes are protected to allow reusability by inheritance.
ATTRIBUTE
ATTRIBUTE is the static constant string for the ARFF keyword '@attribute'.
|
Type
|
String
|
|
Default Value
|
"@attribute"
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
DATA
DATA is the static constant string for the ARFF keyword '@data'. It defines the beginning of the
data block in the ARFF file.
|
Type
|
String
|
|
Default Value
|
"@data"
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
RELATION
RELATION is the static constant with the ARFF keyword '@relation'. It represents the beginning of the ARFF dataset definition.
|
Type
|
String
|
|
Default Value
|
"@relation"
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
addAllValues
This method reads the DATA block in the dataset and adds the values in the file to the corresponding
column structure.
Parameter:
• sColumnFormat String indicating the sequence of
column types that corresponds to the attribute order of each instance in the
dataset.
o s: Nominal column
o f: Numerical (real) column
o c: Categorical column
o b: Binary column
o d: Date column
o %: Skip this column (do not dump its values to any column)
For example, the string “cbbf%%d” indicates the sequence of
attributes that are read from the dataset and copied in memory to the column structure of the
dataset: a categorical column, two binary columns, and a numerical column. The following two attributes are omitted. Finally, the date attribute is copied.
Exceptions:
• IndexOutOfBoundsException
• IOException
• NotAddedValueException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
ArffDataset
Default constructor with no parameters. No dataset filename is
specified using this
constructor.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
ArffDataset
Constructor with the filename of
the dataset as a parameter.
Parameter:
• sFileName The filename of
the dataset
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFileName : String
|
close
This method closes the ARFF file.
Exception:
• IOException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
obtainMetadata
This method reads the metadata of
an ARFF file. Each attribute specification is
interpreted and, if
required, the column structure is
created in the dataset.
This method reads the metadata block of
the dataset. Parameter:
• sColumnFormat String indicating the sequence of
column types that corresponds to the attribute order of each instance in the
dataset.
o s: Nominal column
o f: Numerical (real) column
o c: Categorical column
o b: Binary column
o d: Date column
o %: Skip this column (do not dump its values to any column)
For example, the code "bbf%c" indicates that two binary columns and a numerical (real) column will be
read. Then, the forth attribute will be skipped and, finally, a categorical column will be read.
Exceptions:
• IOException
• InputMismatchException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
open
This method opens the dataset file using the name passed as a parameter to the constructor.
Exceptions:
• FileNotFoundException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
readDataset
This method opens the dataset, reads the metadata and instances and, finally, closes the dataset file.
Parameter:
• sContentFormat Not considered for ARFF datasets
• sColumnFormat String that specifies the types of columns to be read. Each column type is
represented by one of the
following symbols:
o s: Nominal column
o f: Numerical column
o c: Categorical column
o b: Binary column
o d: Date column
o %: Skip this column
Exceptions:
• NotAddedValueException
• IOException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
• inout sContentFormat : String
|
readDataset
This method opens the dataset, reads the metadata and instances and, finally, closes the dataset file.
Parameter:
• sColumnFormat String that specifies the types of columns to be read. Each column type is
represented by one of the
following symbols:
o s: Nominal column
o f: Numerical column
o c: Categorical column
o b: Binary column
o d: Date column
o %: Skip this column
Exceptions:
• NotAddedValueException
• IOException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
readDataset
This method opens the dataset, reads the metadata and instances and, finally, closes the dataset file. The value of
the
column format string is null.
Exceptions:
• NotAddedValueException
• IOException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
writeDataset
This method opens the dataset file, writes metadata and instances, and closes the file. The column
types accepted (otherwise, an InputMismatchException exception is thrown) are the following:
• Numerical
• Date
• Nominal
• Categorical
• Boolean (binary values are saved as categorical values)
Parameter:
• sOutputFile The filename of
the dataset
Exceptions:
• InputMismatchException
• IOException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sOutputFile : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• FileDataset
|
Class
CsvDataset
CsvDataset implements the CSV (Comma-Separated Values) dataset file specification, as prescribed by the
IETF specification, available from http://tools.ietf.org/html/rfc4180
(October, 2005).
Figure
40. Class CsvDataset
|
Name
|
CsvDataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Source::CsvDataset
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• FileDataset
|
|
Realized Interface
|
|
Operation Detail
addAllValues
This method adds all the values in the file to
the corresponding column structure.
Parameter:
• sColumnFormat String indicating the sequence of
column types that corresponds to the attribute order of each instance in the
dataset.
o s: Nominal column
o f: Numerical (real) column
o i: Integer column
o c: Categorical column
o %: Skip this column (do not dump its values to any column)
For example, the string “cf%%s”
indicates the sequence of attributes that are read from the dataset and
copied in memory to the
column structure of
the dataset: a categorical column and a numerical column. The following two attributes are omitted, and, finally, the date attribute is
copied.
Exceptions:
• IndexOutOfBoundsException
• IOException
• NotAddedValueException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
close
This method closes the CSV file.
Exception:
• IOException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
CsvDataset
The default constructor of the CSV dataset with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
CsvDataset
Constructor of the
CSV dataset with its filename as a parameter.
Parameter:
• sFileName The filename of
the CVS dataset
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFileName : String
|
obtainMetadata
This method reads the metadata of
the CSV file. Notice that any metainformation in CSV files is optional.
Parameter:
• sContentFormat String that specifies the structure of the CSV file. The following symbols
are
used:
o n: Indicates that a line with the attribute names is
read
o v: Indicates the block containing the instance values is read
o %: Skip one row in
the file
• sColumnFormat String that specifies the different column types to be read from the file. Each column type is represented by one of
the following symbols:
o s: Nominal column
o f: Numerical (real) column
o c: Categorical column
o i: Integer column
o %: Skip this column
Exceptions:
• IOException
• IllegalFormatSpecificationException
|
Type
|
void
|
|
Visibility
|
Protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
• inout sContentFormat : String
|
open
This method opens the dataset CSV file using the name passed as a parameter to the constructor.
Exceptions:
• FileNotFoundException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
readDataset
This method opens the dataset, reads the metadata and instances and, finally, closes the dataset file.
Parameter:
• sContentFormat String that specifies the structure of the CSV file. The following symbols
are
used:
o n: Indicates that a line with the attribute names is
read
o v: Indicates the block containing the instance values is read
o %: Skip one row in
the file
For example, “%n%%v” omits the first line, then reads the column names, omits
the next two lines and, finally, reads the dataset instances
• sColumnFormat String that specifies the types of columns to be read. Each column type is
represented by one of the
following symbols:
o s: Nominal column
o f: Numerical column
o i: Integer column
o c: Categorical column
o %: Skip this column
Exceptions:
• NotAddedValueException
• IOException
• IndexOutOfBoundsException
• IllegalFormatSpecificationException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
• inout sContentFormat : String
|
readDataset
This method opens the dataset, reads metainformation and instances and, finally, closes the dataset file.
This method assumes the following file format: one first line with the attribute names (metadata),
followed by the instances.
Parameter:
• sColumnFormat String that specifies the types of columns to be read. Each column type is
represented by one of the
following symbols:
o s: Nominal column
o f: Numerical column
o i: Integer column
o c: Categorical column
o %: Skip this column
Exceptions:
• NotAddedValueException
• IOException
• IndexOutOfBoundsException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
writeDataset
This method writes a new CVS dataset file. The column types allowed for writing are the following:
• Numerical
• Integer
• Nominal
• Categorical
• Binary (binary values are saved as categorical values)
Parameter:
• sOutputFile The filename of
the dataset
Exceptions:
• IOException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sOutputFile : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• FileDataset
|
Class
ExcelDataset
ExcelDataset is a class that represents a dataset conformant to the Microsoft Excel standard specification.
This type of
files has the basic features of all spreadsheets, using a grid of
cells arranged in numbered
rows and letter-named columns.
Note: This class has external dependencies to the Java library POI.
Figure
41. Class ExcelDataset
|
Name
|
ExcelDataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Source::ExcelDataset
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• FileDataset
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
addAllValues
This method adds all the values in the DATA block of the file to the corresponding column structure. Parameter:
• sColumnFormat String indicating the sequence of
column types that corresponds to the attribute order of each instance in the
dataset.
o s: Nominal column
o f: Numerical (real) column
o i: Integer column
o c: Categorical column
o %: Skip this column (do not dump its values to any column)
For example, the string “cf%%s” indicates the sequence of attributes that are read from the dataset and
copied in memory to the
column structure of
the dataset: a categorical column and a numerical column. The following two attributes are omitted, and, finally, the nominal attribute is copied.
Exceptions:
• IndexOutOfBoundsException
• IOException
• NotAddedValueException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
close
Close the Excel file.
Exceptions:
• IOException
|
Type
|
void
|
|
Visibility
|
Protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
ExcelDataset
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
ExcelDataset
Constructor with the filename as parameter.
Parameter:
• sFileName The filename of
the Excel dataset
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFileName : String
|
obtainMetadata
This method reads the metadata of
the Excel file.
Parameter:
• sContentFormat String that specifies the data structure in the
Excel file. The following symbols are used:
o n: Indicates that a line with the attribute names is
read
o v: Indicates the block containing the instance values is read
o %: Skip one row in
the file
• sColumnFormat String that specifies the different column types to be read from the file. Each column type is represented by one of
the following symbols:
o s: Nominal column
o f: Numerical (real) column
o c: Categorical column
o i: Integer column
o %: Skip this column
Exceptions:
• IOException
• IllegalFormatSpecificationException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
• inout sContentFormat : String
|
open
This method opens the Excel file using the name passed as a parameter to the constructor.
Exceptions:
• FileNotFoundException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
|
readDataset
This method opens the dataset, reads the metadata and instances and, finally, closes the dataset file.
Parameter:
• sContentFormat String that specifies the structure of the CSV file. The following symbols
are
used:
o n: Indicates that a line with the attribute names is
read
o v: Indicates the block containing the instance values is read
o %: Skip one row in
the file
For example, “%n%%v” omits the first line, then reads the column names, omits
the next two lines and, finally, reads the dataset instances
• sColumnFormat String that specifies the types of columns to be read. Each column type is
represented by one of the
following symbols:
o s: Nominal column
o f: Numerical column
o i: Integer column
o c: Categorical column
o %: Skip this column
Exceptions:
• NotAddedValueException
• IOException
• IndexOutOfBoundsException
• IllegalFormatSpecificationException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
• inout sContentFormat : String
|
writeDataset
This method writes the dataset to
a new Excel file. The column types supported for writing are the
following:
• Numerical
• Integer
• Nominal
• Categorical
• Binary (binary values are saved as categorical values)
Parameter:
• sOutputFile The filename of
the dataset
Exceptions:
• IOException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sOutputFile : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• FileDataset
|
Class
KeelDataset
KeelDataset is the
class representing a dataset conformant to the KEEL (Knowledge Extraction based on
Evolutionary Learning) standard specification. KeelDataset is a subclass of
ArffDataset.
KEEL files are a specific subtype of
ARFF files with the following kind of
attributes:
• @ATTRIBUTE name real [value1,
value2] for real data
• @ATTRIBUTE name integer [value1,
value2] for integer data
• @ATTRIBUTE name {value1,
value2, ...} for categorical data
For a more detailed description of this specification, the reader can consult the following reference:
J. Alcalá-Fdez et al. KEEL Data-Mining Software Tool: Data Set Repository, Integration of Algorithms and
Experimental Analysis Framework. Journal of Multiple-Valued Logic and Soft Computing 17:2-3 (2011) 255-287.
Also, for further information, visit the website http://www.keel.es.
Figure
42. Class KeelDataset
|
Name
|
KeelDataset
|
|
Qualified Name
|
es::uco::kdis::datapro::dataset::Source::KeelDataset
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• ArffDataset
|
|
Realized Interface
|
|
Attribute Detail
INPUTS
Constant for the keyword @inputs
|
Type
|
String
|
|
Default Value
|
"@inputs"
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
OUTPUTS
Constant for the keyword @outputs
|
Type
|
String
|
|
Default Value
|
"@outputs"
|
|
Visibility
|
protected
|
|
Multiplicity
|
|
Operation Detail
addAllValues
This method adds all the values in the @DATA block of
the file to the corresponding column structure.
Parameter:
• sColumnFormat String indicating the sequence of
column types that corresponds to the attribute order of each instance in the
dataset.
o f: Numerical (real) column
o i: Integer column
o c: Categorical column
o b: Binary column
o %: Skip this column (do not dump its values to any column)
For example, the string “cf%%b”
indicates the sequence of attributes that are read from the dataset and
copied in memory to the
column structure of
the dataset: a categorical column and a numerical column. The following two attributes are omitted, and, finally, the binary attribute is
copied.
Exceptions:
• IndexOutOfBoundsException
• IOException
• NotAddedValueException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
KeelDataset
Default constructor with no parameters.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
KeelDataset
Constructor with the filename of
the dataset as a parameter.
Parameter:
• sFileName The filename containing the dataset
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sFileName : String
|
obtainMetadata
This method reads the metadata of
the KEEL file.
Parameter:
• sColumnFormat String that specifies the different column types to be read from the file. Each column type is represented by one of
the following symbols:
o b: Binary column
o f: Numerical (real) column
o c: Categorical column
o i: Integer column
o %: Skip this column
Exceptions:
• IOException
• IllegalFormatSpecificationException
|
Type
|
void
|
|
Visibility
|
protected
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sColumnFormat : String
|
writeDataset
This method writes the dataset to
a new Excel file. Only the following types of column are supported for
writing:
• Numerical (real)
• Integer
• Categorical
Parameter:
• sOutputFile The filename of
the dataset
Exceptions:
• IOException
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout sOutputFile : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• ArffDataset
|
Figure
43. Package es.uco.kdis.datapro.datatypes
|
Name
|
datatypes
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes
|
Class
InvalidValue
This abstract class represents any invalid value in a column. This is
the base class of
the following types of
invalid values:
• Missing values.
• Null values.
• Empty values.
For a more detailed description, see the following reference:
Pyle, D. Data preparation for data mining. Morgan Kaufmann, 1999. ISBN: 1-55869-529-0.
Note. Notice that columns may define their own invalid values. However, these values are not processed by the
library, but only devoted to serialization and specific algorithms. Generally, these objects for invalid values are more than enough for a regular use. Further, these objects are notation-independent, and only
used
for data processing.
Figure
44. Class InvalidValue
|
Name
|
InvalidValue
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes::InvalidValue
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• MissingValue
|
|
Name
|
|
|
Related Element
|
• EmptyValue
|
|
Name
|
|
|
Related Element
|
• NullValue
|
Class
EmptyValue
This class represents an empty value in a variable, i.e., the one for which no real-world value can be
supposed.
This class implements a singleton object, so only one reference can be instantiated simultaneously. Instantiation is
done
using the method getEmptyValue. Therefore, empty values can be compared using
the operator ‘==’.
Figure
45. Class EmptyValue
|
Name
|
EmptyValue
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes::EmptyValue
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• InvalidValue
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
getEmptyValue
Singleton constructor for the object representing an empty value.
|
Type
|
EmptyValue
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• InvalidValue
|
Class
MissingValue
This class represents a missing value in
a variable, i.e., the one that has
not been entered into the dataset, but
for which an actual value exists in the real-world in which the measurements were made.
This class implements a singleton object, so only one reference can be instantiated simultaneously. Instantiation is
done
using the method getMissingValue. Therefore, missing values can be compared
using the operator ‘==’.
Figure
46. Class MissingValue
|
Name
|
MissingValue
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes::MissingValue
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• InvalidValue
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
getMissingValue
Singleton constructor for the object representing a missing value.
|
Type
|
MissingValue
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• InvalidValue
|
This class represents an explicit null value in
a variable.
This class implements a singleton object, so only one reference can be simultaneously instantiated.
Instantiation is
done
using the method getNullValue. Therefore, null values can be compared using the operator ‘==’. Its use allows the programmer to replace null values with comparable object instances (e.g.
in
collections, comparisons, etc.).
Figure
47. Class NullValue
|
Name
|
NullValue
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes::NullValue
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• InvalidValue
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
getNullValue
Singleton constructor for the object representing a null value.
|
Type
|
NullValue
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• InvalidValue
|
This class is
a template to represent any kind of interval consisting of a maximum and minimum limit. These boundaries can be open or close, indicating that the value is
excluded or included in the range. The C defined by the template is
the
class of object involved in the range.
Figure
48. Class Range
|
Name
|
Range
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes::Range
|
|
Visibility
|
public
|
|
Abstract
|
true
|
|
Base Classifier
|
|
|
Realized Interface
|
|
Attribute Detail
Protected attributes with accessors (getter/setter) are omitted.
Operation Detail
getMaxValue
This method returns the upper interval boundary value, i.e. the maximum value in the
interval (the
programmer has to check whether the interval is open or close).
|
Type
|
C
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
getMinValue
This method returns the lower interval boundary value, i.e. the minimum value in the interval (the
programmer has to check whether the interval is open or close).
|
Type
|
C
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
isOpenMax
This method returns a boolean value indicating whether the upper interval boundary is open, i.e. the
maximum value is
excluded from the range.
|
Type
|
boolean
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
isOpenMin
This method returns a boolean value indicating whether the lower interval boundary is open, i.e. the
minimum value is
excluded from the range.
|
Type
|
boolean
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
setMaxValue
This method sets the upper interval boundary.
Parameter:
• oMax The new maximum value
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oMax : C
|
setMinValue
This method sets the lower interval boundary.
Parameter:
• oMin The new minimum value
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout oMin : C
|
setOpenMax
This method sets the upper interval boundary to open or close.
Parameter:
• bOpenMax True if open; false if close.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout bOpenMax : boolean
|
setOpenMin
This method sets the lower interval boundary to open or close.
Parameter:
• bOpenMin True if open; false if close.
|
Type
|
void
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout bOpenMin : boolean
|
Relation Detail
Dependency
|
Name
|
|
|
Related Element
|
• Range<Double>
|
Class
DoubleRange
This class is
a specialization of
the template Range, where the template parameter is of type Double.
Figure
49. Class DoubleRange
|
Name
|
DoubleRange
|
|
Qualified Name
|
es::uco::kdis::datapro::datatypes::DoubleRange
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• Range<Double>
|
|
Realized Interface
|
|
Operation Detail
DoubleRange
Default constructor with no parameters. By default, the lower and upper limit boundaries are set to the negative and positive infinite values, respectively.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
DoubleRange
Constructor with parameters.
Parameters:
• dMin The minimum value of
the range, i.e. the lower interval boundary.
• dMax The maximum value of
the range, i.e. the upper interval boundary.
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dMax : double
• in dMin : double
|
hasValue
This method returns true if the value passed as a parameter is a valid value in the interval.
Parameter:
• dValue The value to be checked.
|
Type
|
boolean
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• in dValue : double
|
toString
This method returns the interval in a String format. The output format is as follows:
‘[‘|’(‘ <min>
‘,’ <max> ‘)’|’]’
where square brackets are used for close intervals, and regular brackets indicate an open value.
|
Type
|
String
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• Range<Double>
|
Figure
50. Package es.uco.kdis.datapro.exception
|
Name
|
exception
|
|
Qualified Name
|
es::uco::kdis::datapro::exception
|
Class
IllegalFormatSpecificationException
This class is
the exception indicating that the file format under consideration does not fulfill the expected standards for such a specification.
Figure
51. Class IllegalFormatSpecificationException
|
Name
|
IllegalFormatSpecificationException
|
|
Qualified Name
|
es::uco::kdis::datapro::exception::IllegalFormatSpecificationException
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• Exception
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
IllegalFormatSpecificationException
Constructor with the error message as a parameter.
Parameter:
• string Error message
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout string : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• Exception
|
Class
NoSuchCategoryException
This class is
the exception indicating that a certain element does not belong to the specified category, or that
a category is
not found.
Figure
52. Class NoSuchCategoryException
|
Name
|
NoSuchCategoryException
|
|
Qualified Name
|
es::uco::kdis::datapro::exception::NoSuchCategoryException
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• Exception
|
|
Realized Interface
|
|
Attribute Detail
All attributes are private.
Operation Detail
NoSuchCategoryException
Constructor with the error message as a parameter.
Parameter:
• string Error message
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout string : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• Exception
|
Class
NotAddedValueException
This class is
the exception indicating that a value was not successfully added to the dataset.
Figure
53. Class NotAddedValueException
|
Name
|
NotAddedValueException
|
|
Qualified Name
|
es::uco::kdis::datapro::exception::NotAddedValueException
|
|
Visibility
|
public
|
|
Abstract
|
false
|
|
Base Classifier
|
• Exception
|
|
Realized Interface
|
|
Attribute Detail
All attribute are private.
Operation Detail
NotAddedValueException
Constructor with the error message as a parameter.
Parameter:
• string Error message
|
Type
|
|
|
Visibility
|
public
|
|
Is Abstract
|
false
|
|
Parameter
|
• inout string : String
|
Relation Detail
Generalization
|
Name
|
|
|
Related Element
|
• Exception
|
Project structure
This project is
structured in
three different parts:
1. Column structure.
2. Datasets hierarchy.
3. Strategies.
Column structure
If the programmer wants to develop new columns or adapt an existing one to
his
own requirements, he
should have in mind the strict separation between abstraction and implementation. The former implements
those methods directly devoted to manage the column metainformation and delegates any processing, handling or query related to the column real values to its implementation. For further information, see the Bridge design pattern (http://en.wikipedia.org/wiki/Bridge_pattern).
We recommend the following guidelines for the development of new
columns:
• Column classes should be located in the
package es.uco.kdis.datapro.dataset.Column
• For a given type of column, namely X, the abstraction class will be named XColumn, and its implementation class, XColumnImpl.
• The new column X has to be added to the enumeration ColumnType. This value is
returned by
the column as its type.
• Column implementations should not be directly accessed from any other class than its abstraction.
Datasets hierarchy
The library provides a finite number of dataset implementations (ARFF, Keel, CSV, MySql, ...
and increasing), but its architecture permits the programmer to extend this part to make his own datasets of interest available. Rarely dataset classes are directly inherited from the top Dataset abstract class, but it is advisable to create, use and maintain the correct class hierarchy where common (both structural and
behavioural) properties are defined, for design reasons. For example, ARFF and CSV datasets will inherit
from the common file-based dataset, i.e. the abstract class FileDataset. Their respective classes will only define those properties that are specific to
these kinds of
file, whereas file-specific properties are defined by intermediate abstract classes. Dataset is always the root of
this hierarchy, since this class links the
physical dataset to the
logical column structure.
Some guidelines to be considered:
• Dataset abstract classes for defining common properties are located in the package
es.uco.kdis.datapro.dataset
• Dataset concrete classes are located in the package es.uco.kdis.datapro.dataset.Source
• Dataset classes should be
named with the suffix -“Dataset”, .e.g, CsvDataset.
Apart from the constructor (with or without parameters), the main methods
to pay attention are inherited from
the abstract class Dataset:
• readDataset, which allows the programmer to configure the type of columns to be filled, as well as and the dataset structure.
• writeDataset, which permits the programmer to save current dataset values into the specific
format.
These methods should fulfill the following assumptions:
• When reading, format can vary or contain errors (invalid values, missing or wrong structure, etc.).
• When reading, the original structure (meta-data) of the
dataset should be recalled somehow.
• When writing, the dataset may have been read from a dataset of the same type, or not:
o If the source dataset is
of the same format, the programmer may want to
overwrite or
generate a new dataset. In both
cases, the resulting dataset should maintain the same structure (e.g. column types and meta-data) than the source dataset.
o If the dataset to be written is of a different type than the source dataset (or the same type
with a different structure), the programmer may want to specify the type of columns to be declared in the resulting dataset.
Strategies
Strategies are the core and most scalable element of the library. Strategies implement algorithms on data.
Strategies are independent of a specific dataset, so they can make use of more than one dataset. See
DatasetStrategy in this guide for more information on the methods that should be
implemented.
To implement your own algorithms, the following guidelines should be considered:
• Every algorithm should be
a subclass of DatasetStrategy.
• Algorithms are grouped in
packages from es.uco.kdis.datapro.algorithm
• Only the package es.uco.kdis.datapro.algorithm.base is required by the library. The rest of packages from
es.uco.kdis.datapro.algorithm could be excluded from the programmer’s
distribution. Notice that each specific algorithm package may have its own external dependencies.
Other packages
Apart from the specific packages for columns, datasets and strategies, there are some other relevant
packages to consider that may be extended as well:
• es.uco.kdis.datapro.datatypes, this package implements the auxiliary classes and datatypes used by datapro4j. For example, the classes declaring invalid values, ranges, etc.
• es.uco.kdis.datapro.exception, this package implements the exception classes. The programmer should look for alternative Java common exceptions before implementing his own class
and clutter the library up with unnecessary classes.
Code documentation
Class headings are documented according to the following structure: class description, contact info and history.
/**
* CLASS DESCRIPTION
*
* <p>
* CONTACT INFO:
* <ul>
* <li>Jose Raul Romero,
PhD [jrromero@uco.es]
* <p>{@link http://www.jrromero.net}
* <p><p>
* Knowledge Discovery and Intelligent Systems Research Group (KDIS) <p>
* {@link http://www.uco.es/grupos/kdis}
* </ul>
* <p>
* HISTORY:
* <ul>
* <li> INCLUDE HERE THE LIST OF CHANGES TO THIS SPECIFIC FILE
* </ul>
* <p>
*
@author Jose
Raul Romero (JRR, 0.2, 0.3) EXAMPLE OF AUTHORS, INITIALS, VERSIONS
@author Jose Maria Luna (JML, 0.1)
@version 0.3
*
**/
Each parameter and method should follow the Javadoc notation for documenting the code.
Further, remember include the file license.txt
in
every distribution that includes the library or part of it.
Coding
recommendations
1.
Code should be implemented following the Hungarian
notation.
2.
Code and comments should
be written in English.
TABLE OF CONTENTS
TABLE OF FIGURES
Introduction
