magpie.data

Class MultiPropertyDataset

java.lang.Object

magpie.data.Dataset

magpie.data.MultiPropertyDataset

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, Citable, Commandable, Options, Printable, Savable

Direct Known Subclasses:

CompositionDataset, ElementDataset

public class MultiPropertyDataset
extends Dataset

Dataset that can store multiple properties for each entry. User can define the names of these properties, how many different classes are possible for each property (if any), and which one to use as the class variable. If employed with MultiPropertyEntry, it will handle the storage of these properties transparently.

Implemented Commands:

target <name> [-keep] - Set class variable to be a certain property
name: Name of property to use as class variable
-keep: Whether to keep entries without a measurement for this property

Implemented Save Formats:

prop: Print out the measured and predicted properties

See Also:

MultiPropertyEntry, Serialized Form

Field Summary

Fields inherited from class magpie.data.Dataset

AttributeName, Entries, Expanders, Generators

Constructor Summary

Constructors

Constructor and Description
MultiPropertyDataset()

Method Summary

Modifier and Type	Method and Description
void	addEntries(java.util.Collection<? extends BaseEntry> entries) Add many entries to a the data set
void	addEntries(Dataset otherDataset, boolean forceMerge) Add entries from another dataset.
void	addEntry(BaseEntry e) Add an entry.
void	addProperty(java.lang.String name) Add a new property to dataset.
void	addProperty(java.lang.String name, java.lang.String[] possibleClasses) Add a new property to dataset.
void	clearPropertyData() Clear all property information from this dataset.
void	combine(Dataset d) Combine the dataset with another.
MultiPropertyDataset	createTemplate() Create a template designed to be saved using serialization
void	definePropertyNames(java.lang.String[] names) Define all known properties for a dataset.
Dataset	emptyClone() Creates a new instance with the same class and attribute names, but without any entries.
java.lang.String[]	getClassNames()
MultiPropertyEntry	getEntry(int index) Retrieve a single entry from the dataset
double[]	getMeasuredPropertyArray(int index) Get the measured value for a certain property for all entries in the dataset.
double[]	getMeasuredPropertyArray(java.lang.String PropertyName) Get measured values for a certain property for all entries in the dataset.
double[]	getPredictedPropertyArray(int index) Get the predicted value for a certain property for all entries in the dataset.
double[]	getPredictedPropertyArray(java.lang.String PropertyName) Get predicted values for a certain property for all entries in the dataset.
int	getPropertyClassCount(int index) Get the number of classes for a certain property
java.lang.String[]	getPropertyClasses(int index) Get list of possible classes for a certain property
java.lang.String[]	getPropertyClasses(java.lang.String name) Get list of possible classes for a certain property
int	getPropertyIndex(java.lang.String name) Lookup the index of a property.
java.lang.String	getPropertyName(int index) Retrieve the name of a certain property.
java.lang.String[]	getPropertyNames() Get the names of each property stored in this dataset.
int	getTargetPropertyIndex() Get the index of the currently active property.
java.lang.String	getTargetPropertyName() Lookup what target property is being used.
boolean	hasProperty(java.lang.String name) Check whether dataset contains a certain property
protected double[]	importEntryProperties(java.lang.String[] words) Used by Dataset.importText(java.lang.String, java.lang.Object[]) to import property measurements for each entry.
protected void	importPropertyNames(java.lang.String line) Given the line describing property names in the input file, read in property names and possible classes.
int	NClasses()
int	NProperties()
java.lang.Object	runCommand(java.util.List<java.lang.Object> Command) Process some command described by a list of Objects.
java.lang.String	saveCommand(java.lang.String Basename, java.lang.String Format) Handles complicated saving commands.
void	setClassNames(java.lang.String[] newClassNames) Set name of class variable (or possible classes)
void	setPropertyClasses(int index, java.lang.String[] classes) Define possible classes for a certain property (like "Metal" or "Nonmetal")
void	setTargetProperty(int index, boolean keepUnmeasured) Define which property to set as the measured class of each entry.
void	setTargetProperty(java.lang.String Property, boolean keepUnmeasured) Define which property to use as the class variable.
org.json.JSONObject	toJSON() Print dataset to JSON format
java.lang.String	toString()
boolean	usingPropertyAsClass()

Methods inherited from class magpie.data.Dataset

about, addAttribute, addAttributeExpander, addAttributeGenerator, addAttributes, addEntry, calculateAttributes, clearAttributeExpanders, clearAttributeGenerators, clearAttributes, clearData, clone, combine, combine, containsEntry, deleteMeasuredClasses, deletePredictedClasses, finalizeGeneration, generateAttributes, getAttributeArray, getAttributeExpanders, getAttributeGenerators, getAttributeIndex, getAttributeName, getAttributeNames, getCitations, getClassName, getClassProbabilityArray, getDistribution, getDistributionCount, getDuplicates, getEntries, getEntriesWriteAccess, getEntryArray, getMeasuredClassArray, getPredictedClassArray, getRandomSplit, getRandomSplit, getRandomSplit, getRandomSplit, getRandomSubset, getRandomSubset, getSingleAttributeArray, getSubset, getTrainingExamples, getUniqueEntries, importText, loadState, matchEntries, NAttributes, NEntries, partition, partition, printAttributeDescription, printCommand, printDescription, printDistribution, printEntries, printEntryDescription, printUsage, removeDuplicates, resolveDuplicates, restoreAttributes, runAttributeCommand, runAttributeExpanders, runAttributeExpansionCommand, runAttributeGeneratorCommand, runAttributeGenerators, saveState, saveTemplate, setAttributeNames, setClassProbabilities, setMeasuredClasses, setOptions, setPredictedClasses, splitForThreading, splitIntoFolds, splitIntoFolds, subtract, transferToWeka

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

MultiPropertyDataset

public MultiPropertyDataset()

Method Detail

emptyClone

public Dataset emptyClone()

Description copied from class: Dataset

Creates a new instance with the same class and attribute names, but without any entries.

Overrides:

emptyClone in class Dataset

Returns:

Dataset with same properties, no entries

createTemplate

public MultiPropertyDataset createTemplate()

Description copied from class: Dataset

Create a template designed to be saved using serialization

Overrides:

createTemplate in class Dataset

Returns:

An empty clone of the dataset, with any additional changes made

getEntry

public MultiPropertyEntry getEntry(int index)

Description copied from class: Dataset

Retrieve a single entry from the dataset

Overrides:

getEntry in class Dataset

Parameters:

index - Index of entry

Returns:

Specified entry

NClasses

public int NClasses()

Overrides:

NClasses in class Dataset

Returns:

Number of possible (discrete) values for class variable. 1 means variable is continuous

addEntries

public void addEntries(Dataset otherDataset,
                       boolean forceMerge)
                throws java.lang.Exception

Add entries from another dataset. If other dataset has different attributes or properties, you can force a merge between these datasets by deleting the attributes and merging the properties.

Entries will be cloned before adding them to this dataset.

For datasets with different properties, the final dataset will contain properties from both datasets. Entries from this datasets that lack properties from the other dataset will have no measured or predicted values for those properties. And, for entries from the other dataset that lack entries from this dataset they will have no measured or predicted values for those properties.

Overrides:

addEntries in class Dataset

Parameters:

otherDataset - Dataset to be added to this one

forceMerge - Whether to force merge if attributes / class / properties are different.

Throws:

java.lang.Exception - If datasets have different classes or attributes, and forceMerge is false.

combine

public void combine(Dataset d)

Combine the dataset with another. Will change the target property of the other dataset to be the same as this one

Overrides:

combine in class Dataset

Parameters:

d - Other dataset. Will change target property to be the same as this class

addEntry

public void addEntry(BaseEntry e)

Description copied from class: Dataset

Add an entry. You may need to run Dataset.generateAttributes()

Overrides:

addEntry in class Dataset

Parameters:

e - Entry to be added

addEntries

public void addEntries(java.util.Collection<? extends BaseEntry> entries)

Description copied from class: Dataset

Add many entries to a the data set

Overrides:

addEntries in class Dataset

Parameters:

entries - Any collection type of entries

NProperties

public int NProperties()

Returns:

Number of properties known by this Dataset

getPropertyNames

public java.lang.String[] getPropertyNames()

Get the names of each property stored in this dataset.

Returns:

Array containing property names

getClassNames

public java.lang.String[] getClassNames()

Overrides:

getClassNames in class Dataset

Returns:

Names of possible classes for class variable

setClassNames

public void setClassNames(java.lang.String[] newClassNames)

Description copied from class: Dataset

Set name of class variable (or possible classes)

Overrides:

setClassNames in class Dataset

Parameters:

newClassNames - New name(s) to use

getPropertyName

public java.lang.String getPropertyName(int index)

Retrieve the name of a certain property.

Parameters:

index - Index of property to look up

Returns:

Name of that property

getTargetPropertyIndex

public int getTargetPropertyIndex()

Get the index of the currently active property. Returns -1 if no property currently in use.

Returns:

Index of target property

usingPropertyAsClass

public boolean usingPropertyAsClass()

Returns:

Whether a property is being used as the class variable.

getTargetPropertyName

public java.lang.String getTargetPropertyName()

Lookup what target property is being used.

Returns:

Name of property being used as class variable. "None" if no property is being used.

getPropertyIndex

public int getPropertyIndex(java.lang.String name)

Lookup the index of a property. Returns -1 if no property by that name exists

Parameters:

name - Name of property

Returns:

Index of that property

setPropertyClasses

public void setPropertyClasses(int index,
                               java.lang.String[] classes)

Define possible classes for a certain property (like "Metal" or "Nonmetal")

Parameters:

index - Index of property

classes - Possible classes

getPropertyClasses

public java.lang.String[] getPropertyClasses(java.lang.String name)

Get list of possible classes for a certain property

Parameters:

name - Name of property

Returns:

List of classes for that property

getPropertyClasses

public java.lang.String[] getPropertyClasses(int index)

Get list of possible classes for a certain property

Parameters:

index - Index of property

Returns:

List of classes for that property

getPropertyClassCount

public int getPropertyClassCount(int index)

Get the number of classes for a certain property

Parameters:

index - Index of property

Returns:

Number of classes for that property

hasProperty

public boolean hasProperty(java.lang.String name)

Check whether dataset contains a certain property

Parameters:

name - Name of property

Returns:

Whether that property is known

definePropertyNames

public void definePropertyNames(java.lang.String[] names)

Define all known properties for a dataset. Removes duplicates

Note: This does not mean that the entries will recognize all of these properties. You must run either MultiPropertyEntry.addProperty(double) or MultiPropertyEntry.setMeasuredProperties(double[]).

Note: Order of properties may be changed

Parameters:

names - Names of properties

setTargetProperty

public void setTargetProperty(java.lang.String Property,
                              boolean keepUnmeasured)

Define which property to use as the class variable.

Parameters:

Property - Name of this property

keepUnmeasured - Whether to keep entries without a measurement for this property

setTargetProperty

public void setTargetProperty(int index,
                              boolean keepUnmeasured)

Define which property to set as the measured class of each entry. Automatically set this as the target property for each entry. Set to "-1" to use a variable besides a measured property as the class variable.

Parameters:

index - Index of the property to be used

keepUnmeasured - Whether to keep entries without a measurement for this property

clearPropertyData

public void clearPropertyData()

Clear all property information from this dataset.

addProperty

public void addProperty(java.lang.String name)

Add a new property to dataset.

NOTE: You must add properties to each entry in the Dataset. This is not done automatically because there are several possible ways to add properties to an entry, depending on what you already know about the entry (see below)

Parameters:

name - Name of new property

See Also:

MultiPropertyEntry.addProperty(), MultiPropertyEntry.addProperty(double), MultiPropertyEntry.addProperty(double, double)

addProperty

public void addProperty(java.lang.String name,
                        java.lang.String[] possibleClasses)

Add a new property to dataset.

NOTE: You must add properties to each entry in the Dataset. This is not done automatically because there are several possible ways to add properties to an entry, depending on what you already know about the entry (see below)

Parameters:

name - Name of new property

possibleClasses - Possible classes for this property

See Also:

MultiPropertyEntry.addProperty(), MultiPropertyEntry.addProperty(double), MultiPropertyEntry.addProperty(double, double)

getMeasuredPropertyArray

public double[] getMeasuredPropertyArray(java.lang.String PropertyName)

Get measured values for a certain property for all entries in the dataset. For entries without a measurement for that property, will return NaN.

Parameters:

PropertyName - Name of desired property

Returns:

Array containing measured property for each entry.

getPredictedPropertyArray

public double[] getPredictedPropertyArray(java.lang.String PropertyName)

Get predicted values for a certain property for all entries in the dataset. For entries without a prediction for that property, will return NaN.

Parameters:

PropertyName - Name of desired property

Returns:

Array containing predicted property for each entry.

getMeasuredPropertyArray

public double[] getMeasuredPropertyArray(int index)

Get the measured value for a certain property for all entries in the dataset. For entries without a measurement for that property, will return NaN.

Parameters:

index - Index of the desired property

Returns:

Array containing measured property for each entry.

getPredictedPropertyArray

public double[] getPredictedPropertyArray(int index)

Get the predicted value for a certain property for all entries in the dataset. For entries without a measurement for that property, will return NaN.

Parameters:

index - Index of the desired property

Returns:

Array containing predicted property for each entry.

toString

public java.lang.String toString()

Overrides:

toString in class Dataset

runCommand

public java.lang.Object runCommand(java.util.List<java.lang.Object> Command)
                            throws java.lang.Exception

Description copied from interface: Commandable

Process some command described by a list of Objects.

Specified by:

runCommand in interface Commandable

Overrides:

runCommand in class Dataset

Parameters:

Command - Command as a list of objects

Returns:

Any output (null if no output generated by command)

Throws:

java.lang.Exception - If something goes wrong

saveCommand

public java.lang.String saveCommand(java.lang.String Basename,
                                    java.lang.String Format)
                             throws java.lang.Exception

Description copied from interface: Savable

Handles complicated saving commands.

Dev Note: Make sure to add save format to Javadoc. See Dataset as an example. Required format:

<save><p><b>format<b> - Description
<br>Optional room to talk more about format </save>

Specified by:

saveCommand in interface Savable

Overrides:

saveCommand in class Dataset

Parameters:

Basename - Name of file without extension

Format - Command specifying format in which to print

Returns:

Filename Path to output file

Throws:

java.lang.Exception - If command not understood

importPropertyNames

protected void importPropertyNames(java.lang.String line)

Given the line describing property names in the input file, read in property names and possible classes.

Parameters:

line - Line describing property names

importEntryProperties

protected double[] importEntryProperties(java.lang.String[] words)

Used by Dataset.importText(java.lang.String, java.lang.Object[]) to import property measurements for each entry.

Parameters:

words - Line describing entry, split into words

Returns:

Property measurements for this entry

toJSON

public org.json.JSONObject toJSON()

Description copied from class: Dataset

Print dataset to JSON format

Overrides:

toJSON in class Dataset

Returns:

JSONObject representing dataset.