org.knime.base.node.mine.regression.logistic.learner3

Class LogisticRegressionContent

java.lang.Object

org.knime.base.node.mine.regression.logistic.learner3.LogisticRegressionContent

final class LogisticRegressionContent
extends Object

Utility class that stores results of logistic regression models. It is used by the learner node model and the predictor node model.

Author:

Heiko Hofer

Constructor Summary

Constructors

Constructor and Description
LogisticRegressionContent(PMMLPortObjectSpec outSpec, List<String> factorList, List<String> covariateList, DataCell targetReferenceCategory, boolean sortTargetCategories, boolean sortFactorsCategories, Matrix beta, double loglike, Matrix covMat, int iter) Create new instance.
LogisticRegressionContent(PMMLPortObjectSpec outSpec, List<String> factorList, List<String> covariateList, Map<? extends String,? extends Integer> vectorLengths, DataCell targetReferenceCategory, boolean sortTargetCategories, boolean sortFactorsCategories, Matrix beta, double loglike, Matrix covMat, int iter) Create new instance.

Method Summary

Modifier and Type	Method and Description
PMMLGeneralRegressionContent	createGeneralRegressionContent() Creates a new PMML General Regression Content from this logistic regression model.
BufferedDataTable	createTablePortObject(ExecutionContext exec) Creates a BufferedDataTable with the
Map<String,Double>	getCoefficients(DataCell logit) Returns the parameters mapped to the coefficients for the given logit.
double	getEstimatedLikelihood()
double	getIntercept(DataCell logit) Returns the value of the intercept for the given logit.
double	getInterceptPValue(DataCell logit) Returns the value of the intercept's p-value.
double	getInterceptStdErr(DataCell logit) Returns the value of the intercept's standard error for the given logit.
double	getInterceptZScore(DataCell logit) Returns the value of the intercept's z-score for the given logit.
int	getIterationCount()
List<DataCell>	getLogits() Logits are elements of the target domain values except of the last one.
List<String>	getParameters() Returns the parameters.
Map<String,Double>	getPValues(DataCell logit) Returns the parameters mapped to the p-value for the given logit.
PMMLPortObjectSpec	getSpec() Returns the spec of the output.
Map<String,Double>	getStandardErrors(DataCell logit) Returns the parameters mapped to the standard error for the given logit.
Map<String,Double>	getZScores(DataCell logit) Returns the parameters mapped to the z-score for the given logit.
(package private) static LogisticRegressionContent	load(ModelContentRO parContent, DataTableSpec spec)
(package private) void	save(ModelContentWO parContent) Save internals to the given content.

Methods inherited from class java.lang.Object

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

Constructor Detail

LogisticRegressionContent

LogisticRegressionContent(PMMLPortObjectSpec outSpec,
                          List<String> factorList,
                          List<String> covariateList,
                          DataCell targetReferenceCategory,
                          boolean sortTargetCategories,
                          boolean sortFactorsCategories,
                          Matrix beta,
                          double loglike,
                          Matrix covMat,
                          int iter)

Create new instance.

Parameters:

outSpec - the spec of the output

factorList - the factors (nominal parameters)

covariateList - the covariates (numeric parameters)

targetReferenceCategory - the target reference category, if not set it is the last category

sortTargetCategories - true when target categories should be sorted

sortFactorsCategories - true when categories of nominal data in the include list should be sorted

beta - the estimated regression factors

loglike - the estimated likelihood

covMat - the covariance matrix

iter - the number of iterations

LogisticRegressionContent

LogisticRegressionContent(PMMLPortObjectSpec outSpec,
                          List<String> factorList,
                          List<String> covariateList,
                          Map<? extends String,? extends Integer> vectorLengths,
                          DataCell targetReferenceCategory,
                          boolean sortTargetCategories,
                          boolean sortFactorsCategories,
                          Matrix beta,
                          double loglike,
                          Matrix covMat,
                          int iter)

Create new instance.

Parameters:

outSpec - the spec of the output

factorList - the factors (nominal parameters)

covariateList - the covariates (numeric parameters)

vectorLengths - the length of vector columns

targetReferenceCategory - the target reference category, if not set it is the last category

sortTargetCategories - true when target categories should be sorted

sortFactorsCategories - true when categories of nominal data in the include list should be sorted

beta - the estimated regression factors

loglike - the estimated likelihood

covMat - the covariance matrix

iter - the number of iterations

Method Detail

getEstimatedLikelihood

public double getEstimatedLikelihood()

Returns:

the likelihood

getIterationCount

public int getIterationCount()

Returns:

the number of irls iterations

getLogits

public List<DataCell> getLogits()

Logits are elements of the target domain values except of the last one.

Returns:

the logits

getParameters

public List<String> getParameters()

Returns the parameters. The follow the notation rule: - for covariate (numeric learning column): "column_name" - for factors (nominal learning columns) there are n-1 entries when n is the number of domain values: "column_name=domain_value"

Returns:

the parameters

getCoefficients

public Map<String,Double> getCoefficients(DataCell logit)

Returns the parameters mapped to the coefficients for the given logit.

Parameters:

logit - the logit

Returns:

the variables mapped to the coefficients

getStandardErrors

public Map<String,Double> getStandardErrors(DataCell logit)

Returns the parameters mapped to the standard error for the given logit.

Parameters:

logit - the logit

Returns:

the parameters mapped to the standard error

getZScores

public Map<String,Double> getZScores(DataCell logit)

Returns the parameters mapped to the z-score for the given logit.

Parameters:

logit - the logit

Returns:

the parameters mapped to the z-score

getPValues

public Map<String,Double> getPValues(DataCell logit)

Returns the parameters mapped to the p-value for the given logit.

Parameters:

logit - the logit

Returns:

the parameters mapped to the p-value

getIntercept

public double getIntercept(DataCell logit)

Returns the value of the intercept for the given logit.

Parameters:

logit - the logit

Returns:

the value of the intercept

getInterceptStdErr

public double getInterceptStdErr(DataCell logit)

Returns the value of the intercept's standard error for the given logit.

Parameters:

logit - the logit

Returns:

the value of the intercept's standard error

getInterceptZScore

public double getInterceptZScore(DataCell logit)

Returns the value of the intercept's z-score for the given logit.

Parameters:

logit - the logit

Returns:

the value of the intercept's z-score

getInterceptPValue

public double getInterceptPValue(DataCell logit)

Returns the value of the intercept's p-value.

Parameters:

logit - the logit

Returns:

the value of the intercept's p-value

createTablePortObject

public BufferedDataTable createTablePortObject(ExecutionContext exec)

Creates a BufferedDataTable with the

Parameters:

exec - The execution context

Returns:

a port object

createGeneralRegressionContent

public PMMLGeneralRegressionContent createGeneralRegressionContent()

Creates a new PMML General Regression Content from this logistic regression model.

Returns:

the PMMLGeneralRegressionContent

load

static LogisticRegressionContent load(ModelContentRO parContent,
                                      DataTableSpec spec)
                               throws InvalidSettingsException

Parameters:

parContent - the content that holds the internals

spec - the data table spec of the training data

Returns:

a instance with he loaded values

Throws:

InvalidSettingsException - when data are not well formed

save

void save(ModelContentWO parContent)

Save internals to the given content.

Parameters:

parContent - the content used as a storage

getSpec

public PMMLPortObjectSpec getSpec()

Returns the spec of the output.

Returns:

spec of the output