edu.cmu.minorthird.classify.algorithms.svm
Class SVMUtils

java.lang.Object
  edu.cmu.minorthird.classify.algorithms.svm.SVMUtils

public class SVMUtils
extends java.lang.Object

Provides some basic utilities for dealing with libsvm. It can convert Features to nodes, instances to node arrays and Datasets to problems.

Author:

ksteppe, Frank Lin

Constructor Summary

SVMUtils()

Method Summary

static libsvm.svm_problem	convertToSVMProblem(Dataset dataset) convert the given dataset into a svm_problem object by looping through the examples and features - features are resorted numericly
static libsvm.svm_node	featureToNode(Feature feature, Instance instance) converts the feature into an svm_node
static libsvm.svm_node[]	instanceToNodeArray(Instance instance) creates the node array from an instance
static Instance	nodeArrayToInstance(libsvm.svm_node[] nodes, FeatureFactory featureFactory) creates an instance from the node array
static Feature	nodeToFeature(libsvm.svm_node node, FeatureFactory featureFactory) convert a svm_node to a feature
static java.lang.String	toString(libsvm.svm_node node)
static java.lang.String	toString(libsvm.svm_node[] nodes)
static java.lang.String	toString(libsvm.svm_problem problem)

Methods inherited from class java.lang.Object

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

Constructor Detail

SVMUtils

public SVMUtils()

Method Detail

toString

public static java.lang.String toString(libsvm.svm_node node)

toString

public static java.lang.String toString(libsvm.svm_node[] nodes)

toString

public static java.lang.String toString(libsvm.svm_problem problem)

featureToNode

public static libsvm.svm_node featureToNode(Feature feature,
                                            Instance instance)

converts the feature into an svm_node

Parameters:

feature - Feature to convert into a node

instance - Instance feature is in - used to retrieve the weight of the feature

Returns:

svm_node

instanceToNodeArray

public static libsvm.svm_node[] instanceToNodeArray(Instance instance)

creates the node array from an instance

Parameters:

instance - Instance to convert

Returns:

node array with all the features from the instance

convertToSVMProblem

public static libsvm.svm_problem convertToSVMProblem(Dataset dataset)

convert the given dataset into a svm_problem object by looping through the examples and features - features are resorted numericly

Parameters:

dataset - - must contain features with integer names

Returns:

a fully loaded svm_problem object

nodeToFeature

public static Feature nodeToFeature(libsvm.svm_node node,
                                    FeatureFactory featureFactory)

convert a svm_node to a feature

Parameters:

node - svm_node from LIBSVM

featureFactory - FeatureIdFactory object holds feature and its id information.

Returns:

Feature Feature converted from svm_node

nodeArrayToInstance

public static Instance nodeArrayToInstance(libsvm.svm_node[] nodes,
                                           FeatureFactory featureFactory)

creates an instance from the node array

Parameters:

nodes - svm_node array from LIBSVM

featureFactory - FeatureFactory object holds feature and its id

Returns:

Instance Instance with the Features converted from input node array