org.apache.ctakes.ytex.kernel

Class FoldGeneratorImpl

java.lang.Object

org.apache.ctakes.ytex.kernel.FoldGeneratorImpl

All Implemented Interfaces:

FoldGenerator

public class FoldGeneratorImpl
extends Object
implements FoldGenerator

utility generates cv fold splits, stores in db. Takes as a command line parameter -prop [property file]. Also reads properties from org.apache.ctakes.ytex.properties. Required properties:

• org.apache.ctakes.ytex.corpusName
• instanceClassQuery

Optional properties:

• minPerClass default 1 minimum number of instances per class/fold. if not enough instances of a specific class, the instance will be repeated across folds. E.g. if you have only one example of one class, and 2 folds, that one example will be duplicated in both folds.
• rand random number seed, defaults to current time millis
• org.apache.ctakes.ytex.splitName default null cv_fold.split_name
• folds default 2
• runs default 5

Author:

vijay

Constructor Summary

Constructors

Constructor and Description
FoldGeneratorImpl()

Method Summary

Modifier and Type	Method and Description
void	generateFolds(Set<String> labels, InstanceData instances, String corpusName, String splitName, int run, String query, int nFolds, int nMinPerClass, Random r) generate folds for a run
SortedMap<String,SortedMap<Integer,SortedMap<Integer,SortedMap<Boolean,SortedMap<Long,String>>>>>	generateRuns(SortedMap<String,SortedMap<Integer,SortedMap<Integer,SortedMap<Boolean,SortedMap<Long,String>>>>> labelToInstanceMap, int nFolds, int nMinPerClass, Integer nSeed, int nRuns) Generate cross validation folds, don't store in database.
void	generateRuns(String corpusName, String splitName, String query, int nFolds, int nMinPerClass, Integer nSeed, int nRuns) Generate cross validation folds, store in database.
ClassifierEvaluationDao	getClassifierEvaluationDao()
KernelUtil	getKernelUtil()
static void	main(String[] args)
void	setClassifierEvaluationDao(ClassifierEvaluationDao classifierEvaluationDao)
void	setKernelUtil(KernelUtil kernelUtil)

Methods inherited from class java.lang.Object

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

Constructor Detail

FoldGeneratorImpl

public FoldGeneratorImpl()

Method Detail

main

public static void main(String[] args)
                 throws org.apache.commons.cli.ParseException,
                        IOException

Throws:

org.apache.commons.cli.ParseException

IOException

generateFolds

public void generateFolds(Set<String> labels,
                          InstanceData instances,
                          String corpusName,
                          String splitName,
                          int run,
                          String query,
                          int nFolds,
                          int nMinPerClass,
                          Random r)

generate folds for a run

Parameters:

labels -

mapInstanceToClassLabel -

name -

splitName -

run -

query -

nFolds -

nMinPerClass -

r -

generateRuns

public void generateRuns(String corpusName,
                         String splitName,
                         String query,
                         int nFolds,
                         int nMinPerClass,
                         Integer nSeed,
                         int nRuns)

Description copied from interface: FoldGenerator

Generate cross validation folds, store in database.

Specified by:

generateRuns in interface FoldGenerator

Parameters:

corpusName - class label

query - query to get instance id - label - class triples

nFolds - number of folds to generate

nMinPerClass - minimum number of instances of each class per fold

nSeed - random number seed; if null will be set currentTime in millis

nRuns - number of runs

getClassifierEvaluationDao

public ClassifierEvaluationDao getClassifierEvaluationDao()

getKernelUtil

public KernelUtil getKernelUtil()

setClassifierEvaluationDao

public void setClassifierEvaluationDao(ClassifierEvaluationDao classifierEvaluationDao)

setKernelUtil

public void setKernelUtil(KernelUtil kernelUtil)

generateRuns

public SortedMap<String,SortedMap<Integer,SortedMap<Integer,SortedMap<Boolean,SortedMap<Long,String>>>>> generateRuns(SortedMap<String,SortedMap<Integer,SortedMap<Integer,SortedMap<Boolean,SortedMap<Long,String>>>>> labelToInstanceMap,
                                                                                                                      int nFolds,
                                                                                                                      int nMinPerClass,
                                                                                                                      Integer nSeed,
                                                                                                                      int nRuns)

Description copied from interface: FoldGenerator

Generate cross validation folds, don't store in database.

Specified by:

generateRuns in interface FoldGenerator

Parameters:

labelToInstanceMap - an instance class map without folds @see InstanceData.labelToInstanceMap

nFolds - number of folds

nMinPerClass - minimum instance per class

nSeed - random seed default to System.currentTimeMillis()

nRuns - number of runs