Package com.namsor.oss.classify.bayes

Class NaiveBayesClassifierMapLaplacedImpl

java.lang.Object

com.namsor.oss.classify.bayes.AbstractNaiveBayesImpl

com.namsor.oss.classify.bayes.AbstractNaiveBayesClassifierImpl

com.namsor.oss.classify.bayes.AbstractNaiveBayesClassifierMapImpl

com.namsor.oss.classify.bayes.NaiveBayesClassifierMapLaplacedImpl

All Implemented Interfaces:

INaiveBayesClassifier

public class NaiveBayesClassifierMapLaplacedImpl
extends AbstractNaiveBayesClassifierMapImpl
implements INaiveBayesClassifier

Naive Bayes Classifier with Laplace smoothing and implementation with concurrent ConcurrentHashMap or persistent mapDB. The Laplace smoothing has two variants as per Sample1 and Sample2.

Author:

elian carsenat, NamSor SAS

Field Summary

Fields inherited from class com.namsor.oss.classify.bayes.AbstractNaiveBayesClassifierImpl

orderByProba

Constructor Summary

Constructors

Constructor	Description
NaiveBayesClassifierMapLaplacedImpl(String classifierName, String[] categories)	Create in-memory classifier using ConcurrentHashMap and defaults for Laplace smoothing (ALPHA=1 and VARIANT=false)
NaiveBayesClassifierMapLaplacedImpl(String classifierName, String[] categories, double alpha, boolean variant)	Create a classifier with in-memory ConcurrentHashMap and Laplace parameters
NaiveBayesClassifierMapLaplacedImpl(String classifierName, String[] categories, double alpha, boolean variant, String rootPathWritable)	Create persistent classifier using org.mapdb.HTreeMap and Laplace parameters
NaiveBayesClassifierMapLaplacedImpl(String classifierName, String[] categories, String rootPathWritable)	Create persistent classifier using org.mapdb.HTreeMap and defaults for Laplace smoothing (ALPHA=1 and VARIANT=false)

Method Summary

Modifier and Type	Method	Description
IClassification	classify(Map<String,String> features, boolean explainData)	Predict most probable class, optionally returning the data needed for future explanation.
void	learn(String category, Map<String,String> features, long weight)	Learn from features

Methods inherited from class com.namsor.oss.classify.bayes.AbstractNaiveBayesClassifierMapImpl

dbClose, dbCloseAndDestroy, dbSize, dumpDb, getDb, getRootPathWritable

Methods inherited from class com.namsor.oss.classify.bayes.AbstractNaiveBayesClassifierImpl

getCategories, getClassifierName, learn, likelihoodsToProbas

Methods inherited from class com.namsor.oss.classify.bayes.AbstractNaiveBayesImpl

pathCategory, pathCategoryFeatureKey, pathCategoryFeatureKeyValue, pathFeatureKey, pathFeatureKeyCountValueTypes, pathFeatureKeyValue, pathGlobal, pathGlobalCountCategories

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.namsor.oss.classify.bayes.INaiveBayesClassifier

dbClose, dbCloseAndDestroy, dbSize, dumpDb, getCategories, learn

Constructor Detail

NaiveBayesClassifierMapLaplacedImpl

public NaiveBayesClassifierMapLaplacedImpl(String classifierName,
                                           String[] categories)

Create in-memory classifier using ConcurrentHashMap and defaults for Laplace smoothing (ALPHA=1 and VARIANT=false)

Parameters:

classifierName - The classifier name

categories - The classification categories

NaiveBayesClassifierMapLaplacedImpl

public NaiveBayesClassifierMapLaplacedImpl(String classifierName,
                                           String[] categories,
                                           double alpha,
                                           boolean variant)

Create a classifier with in-memory ConcurrentHashMap and Laplace parameters

Parameters:

classifierName - The classifier name

categories - The classification categories

alpha - The Laplace alpha, typically 1.0

variant - The Laplace variant

NaiveBayesClassifierMapLaplacedImpl

public NaiveBayesClassifierMapLaplacedImpl(String classifierName,
                                           String[] categories,
                                           double alpha,
                                           boolean variant,
                                           String rootPathWritable)

Create persistent classifier using org.mapdb.HTreeMap and Laplace parameters

Parameters:

classifierName - The classifier name

categories - The classification categories

alpha - The Laplace alpha, typically 1.0

variant - The Laplace variant

rootPathWritable - A writable directory for org.mapdb.HTreeMap storage

NaiveBayesClassifierMapLaplacedImpl

public NaiveBayesClassifierMapLaplacedImpl(String classifierName,
                                           String[] categories,
                                           String rootPathWritable)

Create persistent classifier using org.mapdb.HTreeMap and defaults for Laplace smoothing (ALPHA=1 and VARIANT=false)

Parameters:

classifierName - The classifier name

categories - The classification categories

rootPathWritable - A writable directory for org.mapdb.HTreeMap storage

Method Detail

learn

public void learn(String category,
                  Map<String,String> features,
                  long weight)
           throws ClassifyException

Description copied from interface: INaiveBayesClassifier

Learn from features

Specified by:

learn in interface INaiveBayesClassifier

Parameters:

category - The category

features - The features

weight - The weight

Throws:

ClassifyException - The classification error and cause

classify

public IClassification classify(Map<String,String> features,
                                boolean explainData)
                         throws ClassifyException

Description copied from interface: INaiveBayesClassifier

Predict most probable class, optionally returning the data needed for future explanation.

Specified by:

classify in interface INaiveBayesClassifier

Parameters:

features - The features

explainData - If should return the data needed for future explanation

Returns:

The most likely classes with probability and (optionally) the explanation

Throws:

ClassifyException - The classification error and cause