semeru/text-code-CodeSummarization · Datasets at Hugging Face

input stringlengths 10 828	output stringlengths 5 107
public void clear error list error list clear	clears the list of errors
protected double inner product double x double y double sum 0 for int i 0 i x length i sum x i y i return sum	return the inner product of x and y x
public double get c return c	gets the complexity constant of the svm
protected void prune linked list integer basis set create phit double phi t array new double basis set size for int j 0 j basis set size j phi t array j phi basis set get j phi t new matrix phi t array create diagonal matrix a a new matrix basis set size basis set size for int j 0 j basis set size j a set j j alpha basis set get j	create pruned versions of all important matrices vectors so that
protected void update sigma matrix sigma inv phi t times phi t transpose sigma inv times equals beta sigma inv plus equals a update the factor secholesky decomposition cd new secholesky decomposition sigma inv get array matrix u cd get ptr times cd get l sigma chol u inverse update sigma sigma sigma chol transpose times sigma chol	update the covariance matrix of the weight posterior distribution sigma
public void set alpha int pos double alpha alphas pos alpha	sets an alpha value
protected void reestimate alpha int selected basis alpha selected basis s s q q s	reestimate alpha by setting it to the value which maximizes the
protected void include basis int selected basis basis set add integer value of selected basis reestimate alpha selected basis	include a basis function into the model
protected void delete basis int selected basis basis set remove integer value of selected basis alpha selected basis 1 0d	delete a basis function from the model
public string to string return constructive regression rvm	identify the rvm
public string to string return regression rvm	identify the rvm
protected void project to constraint project alphas to match the constraint double alpha sum sum alpha int svcount 0 double alpha int i for i 0 i examples total i alpha alphas i alpha sum alpha if alpha is zero alpha c neg i is zero alpha is zero alpha c pos i is zero svcount if svcount 0 project alpha sum svcount for i 0 i examples total i alpha alphas i if alpha is zero alpha c neg i is zero alpha is zero alpha c pos i is zero alphas i alpha sum	project variables to constraints
public void add warning string message log warning this get name message error list add message	adds a warning message to the error list
public double sigmoid double x return 1 0 1 0 math exp x	sigmoid link function
public string to string return classification rvm	identify the rvm
public performance vector get estimated performance throws operator exception if pattern throw new user error this 912 this cannot calculate leave one out estimation of error for regression tasks double est vector svmpattern get svm get xi alpha estimation get kernel performance vector pv new performance vector pv add criterion new estimated performance xialpha error est vector 0 1 true pv add criterion new estimated performance xialpha precision est vector 1 1 false pv add criterion new estimated performance xialpha recall est vector 2 1 false pv set main criterion name xialpha error return pv	returns the estimated performances of this svm
protected kernel get kernel return kernel	returns the kernel of this svm
protected svminterface get svm return svm	returns the used svm
public boolean should deliver optimization performance return get parameter as boolean parameter return optimization performance	returns the value of the corresponding parameter
protected void init working set calculate sum int i j project to constraint skip kernel calculation as all alphas 0 for i 0 i examples total i sum i 0 at bound i 0 first working set is random j 0 i 0 while i working set size j examples total working set i j if is alpha neg j which alpha i true else which alpha i false i j update working set	initialises the working set
public performance vector get optimization performance double final fitness get fitness svm examples get alphas svm examples get ys kernel performance vector result new performance vector result add criterion new estimated performance svm objective function final fitness 1 false result add criterion new estimated performance no support vectors svm examples get number of support vectors 1 true return result	returns the optimization performance of the best result
public boolean should calculate weights return get parameter as boolean parameter calculate weights	returns true if the user has specified that weights should be calculated
public attribute weights get weights example set example set throws operator exception if get parameter as int parameter kernel type kernel dot throw new user error this 916 this cannot create weights for nonlinear kernel double weights svm get weights attribute weights weight vector new attribute weights int i 0 for attribute attribute example set get attributes weight vector set weight attribute get name weights i return weight vector	returns the weights for all features
public performance vector get optimization performance return optimization get optimization performance	returns the optimization performance of the best result
public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability numerical class return true if lc learner capability weighted examples return true if lc learner capability formula provider return true return false	returns true for numerical attributes binominal classes and numerical
public performance vector evaluate individual double individual double fitness optimization function get fitness individual ys kernel 0 performance vector result new performance vector result add criterion new estimated performance svmopt value fitness 1 false return result	evaluates the individuals of the given population
protected void logln int level string message if param operator null param operator get log log message rapid miner verbosity level 1 else log service get global log message rapid miner verbosity level 1	log the output plus newline
public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability formula provider return true return false	returns true for numerical attributes binominal classes and numerical
public evo svmmodel train throws operator exception optimize return get model get best values ever	trains the svm
public performance vector get optimization performance double best values ever get best values ever double final fitness optimization function get fitness best values ever ys kernel performance vector result new performance vector result add criterion new estimated performance svm objective function final fitness 0 1 false result add criterion new estimated performance no support vectors 1 final fitness 1 1 true return result	delivers the fitness of the best individual as performance vector
public performance vector get optimization performance double best values ever get best values ever double final fitness optimization function get fitness best values ever ys kernel performance vector result new performance vector if final fitness length 1 result add criterion new estimated performance svm objective function final fitness 0 1 false else result add criterion new estimated performance alpha sum final fitness 0 1 false result add criterion new estimated performance svm objective function final fitness 1 1 false if final fitness length 3 result add criterion new estimated performance alpha label sum final fitness 2 1 false return result	delivers the fitness of the best individual as performance vector
public string to string string type null if get operator description null type get operator class name else type get class get name return break point 0 break point 1 name type	returns the name
public void predict svmexamples to predict int i double prediction svmexample s vmexample int size the examples count examples im 04 02 12 int size to predict count examples for i 0 i size i s vmexample to predict get example i prediction predict s vmexample to predict set y i prediction logln 4 prediction generated	predict values on the testset with model
public double predict svmexample s vmexample int i int sv index double sv att double the sum the examples get b double alpha for i 0 i examples total i alpha alphas i if alpha 0 sv index the examples index i sv att the examples atts i the sum alpha the kernel calculate k sv index sv att s vmexample index s vmexample att return the sum	predict a single example
public string create process tree int indent return create process tree indent null null	returns this operators name and class
public string to string return gauss regression gp	identify the gp
public model learn example set example set throws operator exception attribute label example set get attributes get label if label is nominal if label get mapping size 2 throw new user error this 114 get name label get name init neural net example set train example set return new neural net model example set neural net example set get attributes size this min label this max label	learns and returns a model
public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability polynominal class return true if lc learner capability binominal class return true if lc learner capability numerical class return true if lc learner capability weighted examples return true return false	returns true for all types of attributes and numerical and binominal labels
public boolean supports capability learner capability lc if lc learner capability polynominal attributes return true if lc learner capability binominal attributes return true if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability numerical class return true return false	returns true for all types of attributes and numerical and binominal labels
public string create marked process tree int indent string mark operator mark operator return create process tree indent mark operator mark	returns this operators name and class
public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability numerical class return true return false	returns true for all types of attributes and numerical and binominal labels
protected boolean connect input node i int n if super connect input i n return false double new weights new double weights length 1 system arraycopy weights 0 new weights 0 weights length new weights new weights length 1 this random generator next double 0 1d 0 05d weights new weights double new weight changes new double weight changes length 1 system arraycopy weight changes 0 new weight changes 0 weight changes length new weight changes new weight changes length 1 0 weight changes new weight changes return true	overwrites the super method and also adds weight handling
public model learn example set example set throws operator exception parameter params get parameters example set if example set size 2 throw new user error this 110 2 linear reset random linear disable debug output problem problem get problem example set liblinear model model linear train problem params return new fast margin model example set model get parameter as boolean parameter use bias	learns a new svm model with the lib svm package
protected string create process tree int indent string self prefix string child prefix operator mark operator string mark if mark operator null get name equals mark operator get name return indent indent mark length mark self prefix get name apply count get operator class name else return indent indent self prefix get name apply count get operator class name	returns this operators name and class
private boolean deselect attribute with highest coefficient boolean selected attributes double coefficients double standard deviations double class standard deviation throws undefined parameter error double min coefficient get parameter as double parameter min standardized coefficient int attribute2 deselect 1 int coefficient index 0 for int i 0 i selected attributes length i if selected attributes i double standardized coefficient math abs coefficients coefficient index standard deviations i class standard deviation if standardized coefficient min coefficient min coefficient standardized coefficient attribute2 deselect i coefficient index if attribute2 deselect 0 selected attributes attribute2 deselect false return true return false	this method removes the attribute with the highest standardized coefficient
private double get squared error example set example set boolean selected attributes double coefficients boolean use intercept double error 0 iterator example i example set iterator while i has next example example i next double prediction regression prediction example selected attributes coefficients use intercept double diff prediction example get label error diff diff return error	calculates the squared error of a regression model on the training data
private double regression prediction example example boolean selected attributes double coefficients boolean use intercept double prediction 0 int index 0 int counter 0 for attribute attribute example get attributes if selected attributes counter prediction coefficients index example get value attribute index if use intercept prediction coefficients index return prediction	calculates the prediction for the given example
public double confidence intervall double total weight double total positive weight hypothesis hypo double delta if total weight large return conf small m total weight delta else return conf total weight total positive weight hypo delta	calculates the the confidence intervall for a specific hypothesis
public squared double priors int large super priors large	constructs a new squared with the given default probability
public abstract utility double priors int large this priors new double priors length system arraycopy priors 0 this priors 0 2 this large large	constructor for all utilities
public double utility double total weight double total positive weight hypothesis hypo double g hypo get covered weight total weight double p hypo get positive weight hypo get covered weight if hypo get prediction hypothesis positive class return g g p this priors hypothesis positive class else return g g p this priors hypothesis negative class	calculates the utility for the given number of examples positive examples and hypothesis
public double conf double total weight double delta return inverse normal 1 delta 2 2 math sqrt total weight	calculate confidence intervall without a specific rule for instance averaging functions
public double calculate m double delta double epsilon double i 1 perfomance start with step 10000 while conf i delta epsilon 2 0d i i 10000 if i 1 i i 10000 has been executed at least once i i 10000 while conf i delta epsilon 2 0d i return math ceil i	calculates the m value needed for the gss algorithm
public double conf double total weight double delta double inverse normal inverse normal 1 delta 2 return 3 0d 2 0d math sqrt total weight inverse normal total weight math sqrt total weight 4 0d total weight math sqrt total weight math pow inverse normal 2 0d math pow inverse normal 3 0d 8 0d total weight math sqrt total weight	calculate confidence intervall without a specific rule
public double conf double total weight double total positive weight hypothesis hypo double delta return inverse normal 1 delta 2 variance total weight total positive weight hypo	calculate confidence intervall for a specific rule for instance averaging functions
public double confidence intervall double total weight double delta if total weight large return conf small m total weight delta else return conf total weight delta	calculates the the unspecific confidence intervall
private double variance double p double total weight return p 1 p total weight	calculates the variance for a binomial distribution
private string indent int indent string buffer s new string buffer for int i 0 i indent i s append return s to string	returns a whitespace with length indent
public double confidence intervall double total weight double total positive weight hypothesis hypo double delta if hypo get covered weight large return conf small m total weight delta else return conf total weight total positive weight hypo delta	calculates the the confidence intervall for a specific hypothesis
public double conf small m double total weight double delta double term math log 4 0d delta 2 0d total weight return math pow term 1 5d 3 0d term 3 0d math sqrt term	calculate confidence intervall without a specific rule for small m
public linear double priors int large super priors large	constructs a new linear with the given default probability
public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return g conf g conf 1 0d p0	returns an upper bound for the utility of refinements for the given hypothesis
public double utility double total weight double total positive weight hypothesis hypo double g hypo get covered weight total weight double p hypo get positive weight hypo get covered weight if hypo get prediction hypothesis positive class return g p this priors hypothesis positive class else return g p this priors hypothesis negative class	calculates the utility for the given number of examples positive examples and hypothesis
public coverage double priors int large super priors large	constructs a new coverage with the given default probability
public wracc double priors int large super priors large	constructs new wracc with the given default probability
public double conf double total example weight double delta double inverse normal inverse normal 1 delta 4 return inverse normal math sqrt total example weight math pow inverse normal 2 0d 4 0d total example weight	calculate confidence intervall without a specific rule
public double utility double total weight double total positive weight hypothesis hypo return hypo get positive weight total weight	calculates the utility for the given number of examples positive examples and hypothesis
public double variance double total weight double total positive weight hypothesis hypo double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class double mean this utility total weight total positive weight hypo double inner term hypo get positive weight math pow 1 0 p0 mean 2 hypo get covered weight hypo get positive weight math pow 0 0 p0 mean 2 total weight hypo get covered weight math pow 0 0 mean 2 return math sqrt inner term total weight	calculates the empirical variance
public double variance double total weight double total positive weight hypothesis hypo double mean this utility total weight total positive weight hypo double inner term hypo get positive weight math pow 1 0d mean 2 0d total weight hypo get positive weight math pow 0 0d mean 2 0d return math sqrt inner term total weight	calculates the empirical variance
public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return g conf 1 0 p0	returns an upper bound for the utility of refinements for the given hypothesis
private double variance double p double total example weight return p 1 0d p total example weight	calculates the variance for a binomial distribution
public double get upper bound double total weight double total positive weight hypothesis hypo double delta never needed return 1 0d	returns an upper bound for the utility of refinements for the given hypothesis
public double conf small m double total example weight double delta return 3 0d math sqrt math log 4 0d delta 2 0d total example weight	calculate confidence intervall without a specific rule for small m
public double utility double total example weight double total positive weight hypothesis hypo double fp hypo get covered weight hypo get positive weight double tn total example weight total positive weight fp return hypo get positive weight tn total example weight	calculates the utility for the given number of examples positive examples and hypothesis
public binomial double priors int large super priors large	constructs a new binomial with the given default probability
public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return g conf 1 0d p0	returns an upper bound for the utility of refinements for the given hypothesis
public double utility double total weight double total positive weight hypothesis hypo double g hypo get covered weight total weight double p hypo get positive weight hypo get covered weight if hypo get prediction hypothesis positive class return math sqrt g p this priors hypothesis positive class else return math sqrt g p this priors hypothesis negative class	calculates the utility for the given number of examples positive examples and hypothesis
public double variance double total weight double total positive weight hypothesis hypo double fp hypo get covered weight hypo get positive weight double tn total weight total positive weight fp double correct predictions hypo get positive weight tn double mean correct predictions total weight double inner term correct predictions math pow 1 0d mean 2 total weight correct predictions math pow 0 0d mean 2 return math sqrt inner term total weight	calculates the empirical variance
public hypothesis get hypothesis return hypo	returns the stored hypothesis
public double conf double total weight double delta double inverse normal inverse normal 1 delta 4 return math sqrt inverse normal 2 math sqrt total weight inverse normal 2 math sqrt total weight math pow inverse normal 2 math sqrt total weight 1 5d	calculate confidence intervall without a specific rule
public double get utility return utility	returns the stored utility
public double get upper bound double total weight double total positive weight hypothesis hypo double delta hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double util this utility total weight total positive weight h double conf this confidence intervall total weight delta return util conf	returns an upper bound for the utility of refinements for the given hypothesis
public double get confidence return confidence	returns the stored size of the confidence intervall
public double get total positive weight return total positive weight	returns the stored positive weight
public double conf small m double total example weight double delta double term math log 4 delta 2 total example weight return math sqrt term math pow term 0 25 math pow term 0 75	calculate confidence intervall without a specific rule for small m
public double get total weight return total weight	returns the stored total weight
public boolean equals object o if o instanceof result return false result other result result o if other result get hypothesis equals this get hypothesis return true else return false	returns true if the same hypothesis is stored by both results
public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return math sqrt g conf 1 0d p0	returns an upper bound for the utility of refinements for the given hypothesis
public hypothesis clone rule clone new rule this literals this prediction clone set covered weight this get covered weight clone set positive weight this get positive weight return clone	clones the rule with covered and positive weight
public boolean equals object o if o instanceof literal return false literal other literal literal o if this attribute equals other literal attribute this value other literal value return true else return false	returns true if both attributes and both values are equal
public void apply example e if this applicable e if rejection sampling covered weight if int e get label this prediction positive weight else covered weight e get weight if int e get label this prediction positive weight e get weight	applies the rule to the given examples
public string to string string str this attribute get mapping map index value return attribute get name str	returns a string represenation of this literal
public hypothesis attribute regulars attribute l boolean rs boolean create all rejection sampling rs create all hypothesis create all label l	create a new dummy hypothesis to allow calling the init method
public boolean equals object o if o instanceof gssmodel return false gssmodel other model gssmodel o if other model hypothesis equals this hypothesis return true else return false	returns true if the hypothesis contained in the model are equal
public boolean can be refined no literals can be appended if the last literal tests the last attribute if literals literals length 1 get index all literals length 1 return false else return true	returns true only if this hypothesis can still be refined
public attribute get label return label	returns the label
public void reset this covered weight 0 0d this positive weight 0 0d	sets covered weight and positive weight back to 0
public int get prediction index if this confidences hypothesis positive class this confidences hypothesis negative class return hypothesis positive class else return hypothesis negative class	returns the most probable label index for this model
public double get covered weight return this covered weight	returns the covered weight of this hypothesis
public int get prediction return this prediction	returns the index of prediction of this rule

public void clear error list error list clear

clears the list of errors

protected double inner product double x double y double sum 0 for int i 0 i x length i sum x i y i return sum

return the inner product of x and y x

public double get c return c

gets the complexity constant of the svm

protected void prune linked list integer basis set create phit double phi t array new double basis set size for int j 0 j basis set size j phi t array j phi basis set get j phi t new matrix phi t array create diagonal matrix a a new matrix basis set size basis set size for int j 0 j basis set size j a set j j alpha basis set get j

create pruned versions of all important matrices vectors so that

protected void update sigma matrix sigma inv phi t times phi t transpose sigma inv times equals beta sigma inv plus equals a update the factor secholesky decomposition cd new secholesky decomposition sigma inv get array matrix u cd get ptr times cd get l sigma chol u inverse update sigma sigma sigma chol transpose times sigma chol

update the covariance matrix of the weight posterior distribution sigma

public void set alpha int pos double alpha alphas pos alpha

sets an alpha value

protected void reestimate alpha int selected basis alpha selected basis s s q q s

reestimate alpha by setting it to the value which maximizes the

protected void include basis int selected basis basis set add integer value of selected basis reestimate alpha selected basis

include a basis function into the model

protected void delete basis int selected basis basis set remove integer value of selected basis alpha selected basis 1 0d

delete a basis function from the model

public string to string return constructive regression rvm

identify the rvm

public string to string return regression rvm

identify the rvm

protected void project to constraint project alphas to match the constraint double alpha sum sum alpha int svcount 0 double alpha int i for i 0 i examples total i alpha alphas i alpha sum alpha if alpha is zero alpha c neg i is zero alpha is zero alpha c pos i is zero svcount if svcount 0 project alpha sum svcount for i 0 i examples total i alpha alphas i if alpha is zero alpha c neg i is zero alpha is zero alpha c pos i is zero alphas i alpha sum

project variables to constraints

public void add warning string message log warning this get name message error list add message

adds a warning message to the error list

public double sigmoid double x return 1 0 1 0 math exp x

sigmoid link function

public string to string return classification rvm

identify the rvm

public performance vector get estimated performance throws operator exception if pattern throw new user error this 912 this cannot calculate leave one out estimation of error for regression tasks double est vector svmpattern get svm get xi alpha estimation get kernel performance vector pv new performance vector pv add criterion new estimated performance xialpha error est vector 0 1 true pv add criterion new estimated performance xialpha precision est vector 1 1 false pv add criterion new estimated performance xialpha recall est vector 2 1 false pv set main criterion name xialpha error return pv

returns the estimated performances of this svm

protected kernel get kernel return kernel

returns the kernel of this svm

protected svminterface get svm return svm

returns the used svm

public boolean should deliver optimization performance return get parameter as boolean parameter return optimization performance

returns the value of the corresponding parameter

protected void init working set calculate sum int i j project to constraint skip kernel calculation as all alphas 0 for i 0 i examples total i sum i 0 at bound i 0 first working set is random j 0 i 0 while i working set size j examples total working set i j if is alpha neg j which alpha i true else which alpha i false i j update working set

initialises the working set

public performance vector get optimization performance double final fitness get fitness svm examples get alphas svm examples get ys kernel performance vector result new performance vector result add criterion new estimated performance svm objective function final fitness 1 false result add criterion new estimated performance no support vectors svm examples get number of support vectors 1 true return result

returns the optimization performance of the best result

public boolean should calculate weights return get parameter as boolean parameter calculate weights

returns true if the user has specified that weights should be calculated

public attribute weights get weights example set example set throws operator exception if get parameter as int parameter kernel type kernel dot throw new user error this 916 this cannot create weights for nonlinear kernel double weights svm get weights attribute weights weight vector new attribute weights int i 0 for attribute attribute example set get attributes weight vector set weight attribute get name weights i return weight vector

returns the weights for all features

public performance vector get optimization performance return optimization get optimization performance

returns the optimization performance of the best result

public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability numerical class return true if lc learner capability weighted examples return true if lc learner capability formula provider return true return false

returns true for numerical attributes binominal classes and numerical

public performance vector evaluate individual double individual double fitness optimization function get fitness individual ys kernel 0 performance vector result new performance vector result add criterion new estimated performance svmopt value fitness 1 false return result

evaluates the individuals of the given population

protected void logln int level string message if param operator null param operator get log log message rapid miner verbosity level 1 else log service get global log message rapid miner verbosity level 1

log the output plus newline

public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability formula provider return true return false

returns true for numerical attributes binominal classes and numerical

public evo svmmodel train throws operator exception optimize return get model get best values ever

trains the svm

public performance vector get optimization performance double best values ever get best values ever double final fitness optimization function get fitness best values ever ys kernel performance vector result new performance vector result add criterion new estimated performance svm objective function final fitness 0 1 false result add criterion new estimated performance no support vectors 1 final fitness 1 1 true return result

delivers the fitness of the best individual as performance vector

public performance vector get optimization performance double best values ever get best values ever double final fitness optimization function get fitness best values ever ys kernel performance vector result new performance vector if final fitness length 1 result add criterion new estimated performance svm objective function final fitness 0 1 false else result add criterion new estimated performance alpha sum final fitness 0 1 false result add criterion new estimated performance svm objective function final fitness 1 1 false if final fitness length 3 result add criterion new estimated performance alpha label sum final fitness 2 1 false return result

delivers the fitness of the best individual as performance vector

public string to string string type null if get operator description null type get operator class name else type get class get name return break point 0 break point 1 name type

returns the name

public void predict svmexamples to predict int i double prediction svmexample s vmexample int size the examples count examples im 04 02 12 int size to predict count examples for i 0 i size i s vmexample to predict get example i prediction predict s vmexample to predict set y i prediction logln 4 prediction generated

predict values on the testset with model

public double predict svmexample s vmexample int i int sv index double sv att double the sum the examples get b double alpha for i 0 i examples total i alpha alphas i if alpha 0 sv index the examples index i sv att the examples atts i the sum alpha the kernel calculate k sv index sv att s vmexample index s vmexample att return the sum

predict a single example

public string create process tree int indent return create process tree indent null null

returns this operators name and class

public string to string return gauss regression gp

identify the gp

public model learn example set example set throws operator exception attribute label example set get attributes get label if label is nominal if label get mapping size 2 throw new user error this 114 get name label get name init neural net example set train example set return new neural net model example set neural net example set get attributes size this min label this max label

learns and returns a model

public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability polynominal class return true if lc learner capability binominal class return true if lc learner capability numerical class return true if lc learner capability weighted examples return true return false

returns true for all types of attributes and numerical and binominal labels

public boolean supports capability learner capability lc if lc learner capability polynominal attributes return true if lc learner capability binominal attributes return true if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability numerical class return true return false

returns true for all types of attributes and numerical and binominal labels

public string create marked process tree int indent string mark operator mark operator return create process tree indent mark operator mark

returns this operators name and class

public boolean supports capability learner capability lc if lc learner capability numerical attributes return true if lc learner capability binominal class return true if lc learner capability numerical class return true return false

returns true for all types of attributes and numerical and binominal labels

protected boolean connect input node i int n if super connect input i n return false double new weights new double weights length 1 system arraycopy weights 0 new weights 0 weights length new weights new weights length 1 this random generator next double 0 1d 0 05d weights new weights double new weight changes new double weight changes length 1 system arraycopy weight changes 0 new weight changes 0 weight changes length new weight changes new weight changes length 1 0 weight changes new weight changes return true

overwrites the super method and also adds weight handling

public model learn example set example set throws operator exception parameter params get parameters example set if example set size 2 throw new user error this 110 2 linear reset random linear disable debug output problem problem get problem example set liblinear model model linear train problem params return new fast margin model example set model get parameter as boolean parameter use bias

learns a new svm model with the lib svm package

protected string create process tree int indent string self prefix string child prefix operator mark operator string mark if mark operator null get name equals mark operator get name return indent indent mark length mark self prefix get name apply count get operator class name else return indent indent self prefix get name apply count get operator class name

returns this operators name and class

private boolean deselect attribute with highest coefficient boolean selected attributes double coefficients double standard deviations double class standard deviation throws undefined parameter error double min coefficient get parameter as double parameter min standardized coefficient int attribute2 deselect 1 int coefficient index 0 for int i 0 i selected attributes length i if selected attributes i double standardized coefficient math abs coefficients coefficient index standard deviations i class standard deviation if standardized coefficient min coefficient min coefficient standardized coefficient attribute2 deselect i coefficient index if attribute2 deselect 0 selected attributes attribute2 deselect false return true return false

this method removes the attribute with the highest standardized coefficient

private double get squared error example set example set boolean selected attributes double coefficients boolean use intercept double error 0 iterator example i example set iterator while i has next example example i next double prediction regression prediction example selected attributes coefficients use intercept double diff prediction example get label error diff diff return error

calculates the squared error of a regression model on the training data

private double regression prediction example example boolean selected attributes double coefficients boolean use intercept double prediction 0 int index 0 int counter 0 for attribute attribute example get attributes if selected attributes counter prediction coefficients index example get value attribute index if use intercept prediction coefficients index return prediction

calculates the prediction for the given example

public double confidence intervall double total weight double total positive weight hypothesis hypo double delta if total weight large return conf small m total weight delta else return conf total weight total positive weight hypo delta

calculates the the confidence intervall for a specific hypothesis

public squared double priors int large super priors large

constructs a new squared with the given default probability

public abstract utility double priors int large this priors new double priors length system arraycopy priors 0 this priors 0 2 this large large

constructor for all utilities

public double utility double total weight double total positive weight hypothesis hypo double g hypo get covered weight total weight double p hypo get positive weight hypo get covered weight if hypo get prediction hypothesis positive class return g g p this priors hypothesis positive class else return g g p this priors hypothesis negative class

calculates the utility for the given number of examples positive examples and hypothesis

public double conf double total weight double delta return inverse normal 1 delta 2 2 math sqrt total weight

calculate confidence intervall without a specific rule for instance averaging functions

public double calculate m double delta double epsilon double i 1 perfomance start with step 10000 while conf i delta epsilon 2 0d i i 10000 if i 1 i i 10000 has been executed at least once i i 10000 while conf i delta epsilon 2 0d i return math ceil i

calculates the m value needed for the gss algorithm

public double conf double total weight double delta double inverse normal inverse normal 1 delta 2 return 3 0d 2 0d math sqrt total weight inverse normal total weight math sqrt total weight 4 0d total weight math sqrt total weight math pow inverse normal 2 0d math pow inverse normal 3 0d 8 0d total weight math sqrt total weight

calculate confidence intervall without a specific rule

public double conf double total weight double total positive weight hypothesis hypo double delta return inverse normal 1 delta 2 variance total weight total positive weight hypo

calculate confidence intervall for a specific rule for instance averaging functions

public double confidence intervall double total weight double delta if total weight large return conf small m total weight delta else return conf total weight delta

calculates the the unspecific confidence intervall

private double variance double p double total weight return p 1 p total weight

calculates the variance for a binomial distribution

private string indent int indent string buffer s new string buffer for int i 0 i indent i s append return s to string

returns a whitespace with length indent

public double confidence intervall double total weight double total positive weight hypothesis hypo double delta if hypo get covered weight large return conf small m total weight delta else return conf total weight total positive weight hypo delta

calculates the the confidence intervall for a specific hypothesis

public double conf small m double total weight double delta double term math log 4 0d delta 2 0d total weight return math pow term 1 5d 3 0d term 3 0d math sqrt term

calculate confidence intervall without a specific rule for small m

public linear double priors int large super priors large

constructs a new linear with the given default probability

public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return g conf g conf 1 0d p0

returns an upper bound for the utility of refinements for the given hypothesis

public double utility double total weight double total positive weight hypothesis hypo double g hypo get covered weight total weight double p hypo get positive weight hypo get covered weight if hypo get prediction hypothesis positive class return g p this priors hypothesis positive class else return g p this priors hypothesis negative class

calculates the utility for the given number of examples positive examples and hypothesis

public coverage double priors int large super priors large

constructs a new coverage with the given default probability

public wracc double priors int large super priors large

constructs new wracc with the given default probability

public double conf double total example weight double delta double inverse normal inverse normal 1 delta 4 return inverse normal math sqrt total example weight math pow inverse normal 2 0d 4 0d total example weight

calculate confidence intervall without a specific rule

public double utility double total weight double total positive weight hypothesis hypo return hypo get positive weight total weight

calculates the utility for the given number of examples positive examples and hypothesis

public double variance double total weight double total positive weight hypothesis hypo double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class double mean this utility total weight total positive weight hypo double inner term hypo get positive weight math pow 1 0 p0 mean 2 hypo get covered weight hypo get positive weight math pow 0 0 p0 mean 2 total weight hypo get covered weight math pow 0 0 mean 2 return math sqrt inner term total weight

calculates the empirical variance

public double variance double total weight double total positive weight hypothesis hypo double mean this utility total weight total positive weight hypo double inner term hypo get positive weight math pow 1 0d mean 2 0d total weight hypo get positive weight math pow 0 0d mean 2 0d return math sqrt inner term total weight

calculates the empirical variance

public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return g conf 1 0 p0

returns an upper bound for the utility of refinements for the given hypothesis

private double variance double p double total example weight return p 1 0d p total example weight

calculates the variance for a binomial distribution

public double get upper bound double total weight double total positive weight hypothesis hypo double delta never needed return 1 0d

returns an upper bound for the utility of refinements for the given hypothesis

public double conf small m double total example weight double delta return 3 0d math sqrt math log 4 0d delta 2 0d total example weight

calculate confidence intervall without a specific rule for small m

public double utility double total example weight double total positive weight hypothesis hypo double fp hypo get covered weight hypo get positive weight double tn total example weight total positive weight fp return hypo get positive weight tn total example weight

calculates the utility for the given number of examples positive examples and hypothesis

public binomial double priors int large super priors large

constructs a new binomial with the given default probability

public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return g conf 1 0d p0

returns an upper bound for the utility of refinements for the given hypothesis

public double utility double total weight double total positive weight hypothesis hypo double g hypo get covered weight total weight double p hypo get positive weight hypo get covered weight if hypo get prediction hypothesis positive class return math sqrt g p this priors hypothesis positive class else return math sqrt g p this priors hypothesis negative class

calculates the utility for the given number of examples positive examples and hypothesis

public double variance double total weight double total positive weight hypothesis hypo double fp hypo get covered weight hypo get positive weight double tn total weight total positive weight fp double correct predictions hypo get positive weight tn double mean correct predictions total weight double inner term correct predictions math pow 1 0d mean 2 total weight correct predictions math pow 0 0d mean 2 return math sqrt inner term total weight

calculates the empirical variance

public hypothesis get hypothesis return hypo

returns the stored hypothesis

public double conf double total weight double delta double inverse normal inverse normal 1 delta 4 return math sqrt inverse normal 2 math sqrt total weight inverse normal 2 math sqrt total weight math pow inverse normal 2 math sqrt total weight 1 5d

calculate confidence intervall without a specific rule

public double get utility return utility

returns the stored utility

public double get upper bound double total weight double total positive weight hypothesis hypo double delta hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double util this utility total weight total positive weight h double conf this confidence intervall total weight delta return util conf

returns an upper bound for the utility of refinements for the given hypothesis

public double get confidence return confidence

returns the stored size of the confidence intervall

public double get total positive weight return total positive weight

returns the stored positive weight

public double conf small m double total example weight double delta double term math log 4 delta 2 total example weight return math sqrt term math pow term 0 25 math pow term 0 75

calculate confidence intervall without a specific rule for small m

public double get total weight return total weight

returns the stored total weight

public boolean equals object o if o instanceof result return false result other result result o if other result get hypothesis equals this get hypothesis return true else return false

returns true if the same hypothesis is stored by both results

public double get upper bound double total weight double total positive weight hypothesis hypo double delta double p0 if hypo get prediction hypothesis positive class p0 this priors hypothesis positive class else p0 this priors hypothesis negative class utility cov new coverage this priors this large hypothesis h hypo clone h set covered weight hypo get positive weight all fp become tn double g cov utility total weight total positive weight h double conf cov confidence intervall total weight delta return math sqrt g conf 1 0d p0

returns an upper bound for the utility of refinements for the given hypothesis

public hypothesis clone rule clone new rule this literals this prediction clone set covered weight this get covered weight clone set positive weight this get positive weight return clone

clones the rule with covered and positive weight

public boolean equals object o if o instanceof literal return false literal other literal literal o if this attribute equals other literal attribute this value other literal value return true else return false

returns true if both attributes and both values are equal

public void apply example e if this applicable e if rejection sampling covered weight if int e get label this prediction positive weight else covered weight e get weight if int e get label this prediction positive weight e get weight

applies the rule to the given examples

public string to string string str this attribute get mapping map index value return attribute get name str

returns a string represenation of this literal

public hypothesis attribute regulars attribute l boolean rs boolean create all rejection sampling rs create all hypothesis create all label l

create a new dummy hypothesis to allow calling the init method

public boolean equals object o if o instanceof gssmodel return false gssmodel other model gssmodel o if other model hypothesis equals this hypothesis return true else return false

returns true if the hypothesis contained in the model are equal

public boolean can be refined no literals can be appended if the last literal tests the last attribute if literals literals length 1 get index all literals length 1 return false else return true

returns true only if this hypothesis can still be refined

public attribute get label return label

returns the label

public void reset this covered weight 0 0d this positive weight 0 0d

sets covered weight and positive weight back to 0

public int get prediction index if this confidences hypothesis positive class this confidences hypothesis negative class return hypothesis positive class else return hypothesis negative class

returns the most probable label index for this model

public double get covered weight return this covered weight

returns the covered weight of this hypothesis

public int get prediction return this prediction

returns the index of prediction of this rule