org.knime.base.node.mine.bayes.naivebayes.datamodel
Class NaiveBayesModel

java.lang.Object
  org.knime.base.node.mine.bayes.naivebayes.datamodel.NaiveBayesModel

public class NaiveBayesModel
extends Object

This class represents the learned Naive Bayes model. This basic model holds for each attribute an AttributeModel. Which provides the probability information for each class value.

Author:

Tobias Koetter, University of Konstanz

Field Summary

static NumberFormat

HTML_VALUE_FORMATER The NumberFormater to use in the html views.

Constructor Summary

NaiveBayesModel(BufferedDataTable data, String classColName, ExecutionContext exec, int maxNoOfNominalVals, boolean skipMissingVals)
Constructor which iterates through the DataTable to calculate the needed Bayes variables.

NaiveBayesModel(ConfigRO predParams)
Constructor for class NaiveBayesModel.

Method Summary

List<String>	check4MissingCols(DataTableSpec tableSpec) Checks if the model contains attributes which are not present in the given table specification which could influence the prediction result.
List<String>	check4UnknownCols(DataTableSpec tableSpec) Checks if the given table specification contains columns which are not covered by the learned model.
boolean	containsSkippedAttributes()
AttributeModel	getAttributeModel(String attributeName)
Collection<AttributeModel>	getAttributeModels()
List<String>	getAttributesWithMissingVals()
DataType	getClassColumnDataType()
String	getClassColumnName()
double	getClassPriorProbability(String classValue)
double[]	getClassProbabilities(String[] attributeNames, DataRow row, List<String> classValues, boolean normalize, double laplaceCorrector)
String	getHTMLView()
String	getMostLikelyClass(String[] attrNames, DataRow row, double laplaceCorrector) Returns the name of the class with the highest probability for the given row.
int	getNoOfRecs()
List<AttributeModel>	getSkippedAttributes()
String	getSkippedAttributesString(int max2Show)
List<String>	getSortedClassValues()
String	getSummary()
void	savePredictorParams(ConfigWO predParams)
String	toString()
void	updateModel(DataRow row, DataTableSpec tableSpec, int classColIdx) Updates the current NaiveBayesModel with the values from the given DataRow.

Methods inherited from class java.lang.Object

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

Field Detail

HTML_VALUE_FORMATER

public static final NumberFormat HTML_VALUE_FORMATER

The NumberFormater to use in the html views.

Constructor Detail

NaiveBayesModel

public NaiveBayesModel(BufferedDataTable data,
                       String classColName,
                       ExecutionContext exec,
                       int maxNoOfNominalVals,
                       boolean skipMissingVals)
                throws CanceledExecutionException,
                       InvalidSettingsException

Constructor which iterates through the DataTable to calculate the needed Bayes variables.

Parameters:

data - The BufferedDataTable with the data

classColName - The name of the column with the class

exec - the ExecutionContext to provide progress information and check for cancel

maxNoOfNominalVals - the maximum number of supported unique nominal attribute values

skipMissingVals - set to true if the missing values should be skipped during learning and prediction

Throws:

CanceledExecutionException - if the user presses the cancel button during model creation

InvalidSettingsException - if the input data contains no rows

NaiveBayesModel

public NaiveBayesModel(ConfigRO predParams)
                throws InvalidSettingsException

Constructor for class NaiveBayesModel.

Parameters:

predParams - the ModelContentRO to read from

Throws:

InvalidSettingsException - if a mandatory key is not available

Method Detail

updateModel

public void updateModel(DataRow row,
                        DataTableSpec tableSpec,
                        int classColIdx)
                 throws InvalidSettingsException

Updates the current NaiveBayesModel with the values from the given DataRow.

Parameters:

row - DataRow with values for update

tableSpec - underlying DataTableSpec

classColIdx - the index of the class column

Throws:

InvalidSettingsException - if missing values occur in class column or an attribute has too many values.

savePredictorParams

public void savePredictorParams(ConfigWO predParams)

Parameters:

predParams - to save the model

containsSkippedAttributes

public boolean containsSkippedAttributes()

Returns:

true if the model contains skipped attributes

getSkippedAttributes

public List<AttributeModel> getSkippedAttributes()

Returns:

the skipped attributes or an empty list

getSkippedAttributesString

public String getSkippedAttributesString(int max2Show)

Parameters:

max2Show - the maximum number of missing attributes to display

Returns:

a String that shows the skipped attributes

getSortedClassValues

public List<String> getSortedClassValues()

Returns:

all class values in natural order

getClassPriorProbability

public double getClassPriorProbability(String classValue)

Parameters:

classValue - the value of the class we want the probability for

Returns:

the prior probability for the given class

getClassProbabilities

public double[] getClassProbabilities(String[] attributeNames,
                                      DataRow row,
                                      List<String> classValues,
                                      boolean normalize,
                                      double laplaceCorrector)

Parameters:

attributeNames - the name of the attributes we want the normalized probability values for

row - the row with the values in the same order like the attribute names

classValues - the class values to calculate the probability for

normalize - set to true if the probability values should be normalized

laplaceCorrector - the Laplace corrector to use. A value greater 0 tolerates zero counts (i.e. does not produce 0 probabilities)

Returns:

the probability values in the same order like the class values

getNoOfRecs

public int getNoOfRecs()

Returns:

the total number of training records

getClassColumnName

public String getClassColumnName()

Returns:

the name of the column with the class attribute.

getClassColumnDataType

public DataType getClassColumnDataType()

Returns:

the DataType of the column with the class attribute.

getSummary

public String getSummary()

Returns:

the a summary of the model

getHTMLView

public String getHTMLView()

Returns:

a HTML representation of all attribute models

getAttributesWithMissingVals

public List<String> getAttributesWithMissingVals()

Returns:

the name of all attributes which has at least one missing value during learning or an empty list

getAttributeModel

public AttributeModel getAttributeModel(String attributeName)

Parameters:

attributeName - the name of the attribute

Returns:

the model for the given attribute name or null if the attribute is not known

getAttributeModels

public Collection<AttributeModel> getAttributeModels()

Returns:

an unmodifiable Collection with all AttributeModel objects

getMostLikelyClass

public String getMostLikelyClass(String[] attrNames,
                                 DataRow row,
                                 double laplaceCorrector)

Returns the name of the class with the highest probability for the given row.

Parameters:

attrNames - the attribute names in the same order they appear in the given data row

row - the row with the attributes in the same order like in the training data table

laplaceCorrector - the Laplace corrector to use. A value greater 0 overcomes zero counts

Returns:

the class attribute with the highest probability for the given attribute values.

toString

public String toString()

Overrides:

toString in class Object

check4UnknownCols

public List<String> check4UnknownCols(DataTableSpec tableSpec)

Checks if the given table specification contains columns which are not covered by the learned model. Either because the name is not known or the type is wrong.

Parameters:

tableSpec - the DataTableSpec to check for unknown columns

Returns:

the name of the unknown columns or an empty List

check4MissingCols

public List<String> check4MissingCols(DataTableSpec tableSpec)

Checks if the model contains attributes which are not present in the given table specification which could influence the prediction result.

Parameters:

tableSpec - the DataTableSpec to check for missing columns

Returns:

the name of the missing columns or an empty List