VowpalWabbit.Builder (Vowpal Wabbit Wrapper 1.0.0-SNAPSHOT API)

java.lang.Object
- com.indeed.vw.wrapper.api.VowpalWabbit.Builder

All Implemented Interfaces:

DebuggingOptions, FeatureEngineeringFunctions, FeatureSelectionOptions, LinkAndLossOptions, MiscOptions, OptionsToSaveAndLoadModel, OptionToExchangeRAMForQuality, PredictionBoundaryOptions, RegularizationOptions, SGDVowpalWabbitBuilder, UpdatesOptions

Enclosing class:

VowpalWabbit
```
public static class VowpalWabbit.Builder
extends Object
implements SGDVowpalWabbitBuilder
```
Builder for Vowpal Wabbit learner.

Method Summary

All Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`VowpalWabbit.Builder`	`active()` enable active learning
`VowpalWabbit.Builder`	`activeCover()` enable active learning with cover
`VowpalWabbit.Builder`	`adaptive()` use adaptive, individual learning rates.
`VowpalWabbit.Builder`	`affix(String arg)` generate prefixes/suffixes of features; argument '+2a,-3b,+1'
`VowpalWabbit.Builder`	`auditRegressor(Path regressor)` stores feature names and their regressor values.
`VowpalWabbit.Builder`	`autolink(int d)` create link function with polynomial d
`VowpalWabbit.Builder`	`bfgs()` use bfgs optimization
`VowpalWabbit.Builder`	`binary()` report loss as binary classification on -1,1
`VowpalWabbit.Builder`	`bitPrecision(int bitsNum)` number of bits in the feature table.
`VowpalWabbit.Builder`	`boosting(int n)` Online boosting with <N> weak learners
`VowpalWabbit.Builder`	`bootstrap(int k)` k-way bootstrap by online importance resampling
`com.indeed.vw.wrapper.learner.VWFloatArrayLearner`	`buildFloatArrayLearner()`
`com.indeed.vw.wrapper.learner.VWFloatLearner`	`buildFloatLearner()` Build learner
`com.indeed.vw.wrapper.learner.VWIntArrayLearner`	`buildIntArrayLearner()`
`com.indeed.vw.wrapper.learner.VWIntLearner`	`buildIntLearner()`
`VowpalWabbit.Builder`	`cache()` Use a cache.
`VowpalWabbit.Builder`	`cacheFile(Path cacheFile)` The location(s) of cacheFile.
`VowpalWabbit.Builder`	`cb(int k)` Use contextual bandit learning with <k> costs
`VowpalWabbit.Builder`	`cbAdf()` Do Contextual Bandit learning with multiline action dependent features.
`VowpalWabbit.Builder`	`cbExplore(int k)` Online explore-exploit for a <k> action contextual bandit problem
`VowpalWabbit.Builder`	`cbExploreAdf()` Online explore-exploit for a contextual bandit problem with multiline action dependent features
`VowpalWabbit.Builder`	`cbify(int k)` Convert multiclass on <k> classes into a contextual bandit problem
`VowpalWabbit.Builder`	`compressed()` use gzip format whenever possible.
`VowpalWabbit.Builder`	`confidence()` Get confidence for binary predictions
`VowpalWabbit.Builder`	`confidenceAfterTraining()` Confidence after training
`VowpalWabbit.Builder`	`conjugateGradient()` use conjugate gradient based optimization
`VowpalWabbit.Builder`	`constant(double initialValue)` Set initial value of constant
`VowpalWabbit.Builder`	`csoaa(int k)` One-against-all multiclass with <k> costs
`VowpalWabbit.Builder`	`csoaaLdf(VowpalWabbit.LDF ldf)` Use one-against-all multiclass learning with label dependent features.
`VowpalWabbit.Builder`	`cubic(String firstNameSpace, String secondNamespace, String thirdNamespace)` Create and use cubic features
`VowpalWabbit.Builder`	`decayLearningRate(double decay)` Set Decay factor for learning_rate between passes
`VowpalWabbit.Builder`	`dictionary(Path file)` read a dictionary for additional features (arg either 'x:file' or just 'file')
`VowpalWabbit.Builder`	`dictionaryPath(Path dir)` look in this directory for dictionaries; defaults to current directory or env{PATH}
`VowpalWabbit.Builder`	`earlyTerminate(int passes)` Specify the number of passes tolerated when holdout loss doesn't decrease before early termination, default is 3
`VowpalWabbit.Builder`	`ect(int k)` Error correcting tournament with <k> labels
`VowpalWabbit.Builder`	`examples(int examples)` number of examples to parse
`VowpalWabbit.Builder`	`featureLimit(int n)` limit to N features.
`VowpalWabbit.Builder`	`featureMask(Path featureMask)` Use existing regressor to determine which parameters may be updated.
`VowpalWabbit.Builder`	`finalRegressor(Path regressor)` Final regressor
`VowpalWabbit.Builder`	`ftrl()` FTRL: Follow the Proximal Regularized Leader
`String`	`getCommand()` Get command option will be passes to VWLearner
`VowpalWabbit.Builder`	`hash(VowpalWabbit.Hash hash)` how to hash the features.
`VowpalWabbit.Builder`	`holdoutAfter(int n)` holdout after n training examples, default off (disables holdoutPeriod)
`VowpalWabbit.Builder`	`holdoutPeriod(int holdout)` holdout period for test only, default 10
`VowpalWabbit.Builder`	`id(String id)` User supplied ID embedded into the final regressor
`VowpalWabbit.Builder`	`ignore(String namespace)` ignore namespace <arg>
`VowpalWabbit.Builder`	`initialPassLength(int examples)` initial number of examples per pass
`VowpalWabbit.Builder`	`initialRegressor(Path initialRegressor)` Initial regressor(s)
`VowpalWabbit.Builder`	`initialT(double initialT)` initial t value
`VowpalWabbit.Builder`	`initialWeight(double weight)` Set all weights to an initial value of arg.
`VowpalWabbit.Builder`	`inputFeatureRegularizer(Path regularizationPath)` Per feature regularization input file
`VowpalWabbit.Builder`	`interact(String arg)` Put weights on feature products from namespaces <n1> and <n2>
`VowpalWabbit.Builder`	`interactions(String... namespaces)` Create feature interactions of any level between namespaces.
`VowpalWabbit.Builder`	`invariant()` use safe/importance aware updates.
`VowpalWabbit.Builder`	`keep(String namespace)` keep namespace <arg>
`VowpalWabbit.Builder`	`killCache()` do not reuse existing cache: create a new one always
`VowpalWabbit.Builder`	`ksvm()` kernel svm
`VowpalWabbit.Builder`	`l1(double l1)` l_1 lambda
`VowpalWabbit.Builder`	`l2(double l2)` l_2 lambda
`VowpalWabbit.Builder`	`lda(int topics)` Run lda with <int> topics
`VowpalWabbit.Builder`	`learningRate(double learningRate)` Set learning rate
`VowpalWabbit.Builder`	`leaveDuplicateInteractions()` Don't remove interactions with duplicate combinations of namespaces.
`VowpalWabbit.Builder`	`link(Link link)` Specify the link function: identity, logistic, glf1 or poisson (=identity)
`VowpalWabbit.Builder`	`logMulti(int k)` Use online tree for multiclass
`VowpalWabbit.Builder`	`lossFunction(Loss loss)` Specify the loss function to be used, uses squared by default.
`VowpalWabbit.Builder`	`lrq(String firstNamespace, String secondNamespace, int k)` use low rank quadratic features
`VowpalWabbit.Builder`	`lrqdropout()` use dropout training for low rank quadratic features
`VowpalWabbit.Builder`	`lrqfa(String firstNamespace, String secondNamespace, int k)` use low rank quadratic features with field aware weights
`VowpalWabbit.Builder`	`maxPrediction(double max)` Largest prediction to output
`VowpalWabbit.Builder`	`minPrediction(double min)` Smallest prediction to output
`VowpalWabbit.Builder`	`multilabelOaa(int k)` One-against-all multilabel with <k> labels
`VowpalWabbit.Builder`	`multiworldTest(String arg)` Evaluate features as a policies
`VowpalWabbit.Builder`	`namedLabels(String... labels)` use names for labels (multiclass, etc.) rather than integers, argument specified all possible labels, comma-sep,
`VowpalWabbit.Builder`	`newMf(int rank)` rank for reduction-based matrix factorization
`VowpalWabbit.Builder`	`ngram(String namespace, int n)` Generate N grams.
`VowpalWabbit.Builder`	`nn(int units)` Sigmoidal feedforward network with <k> hidden units
`VowpalWabbit.Builder`	`noconstant()` Don't add a constant feature
`VowpalWabbit.Builder`	`noop()` do no learning
`VowpalWabbit.Builder`	`normalized()` use per feature normalized updates
`VowpalWabbit.Builder`	`oaa(int k)` One-against-all multiclass with <k> labels
`VowpalWabbit.Builder`	`ojaNewton()` Online Newton with Oja's Sketch
`VowpalWabbit.Builder`	`outputFeatureRegularizerBinary(Path regularizationFile)` Per feature regularization output file
`VowpalWabbit.Builder`	`outputFeatureRegularizerText(Path regularizationFile)` Per feature regularization output file, in text
`VowpalWabbit.Builder`	`parameter(String argumentLine)` Add vowpal wabit argument
`VowpalWabbit.Builder`	`passes(int passes)` Number of Training Passes
`VowpalWabbit.Builder`	`permutations()` Use permutations instead of combinations for feature interactions of same namespace.
`VowpalWabbit.Builder`	`pistol()` FTRL: Parameter-free Stochastic Learning
`VowpalWabbit.Builder`	`powerT(double powerT)` t power value
`VowpalWabbit.Builder`	`quadratic(String firstNameSpace, String secondNamespace)` Create and use quadratic features
`VowpalWabbit.Builder`	`quantileTau(double tau)` Parameter \tau associated with Quantile loss.
`VowpalWabbit.Builder`	`randomSeed(int seed)` seed random number generator
`VowpalWabbit.Builder`	`randomWeights(double arg)` make initial weights random
`VowpalWabbit.Builder`	`rank(int rank)` rank for matrix factorization.
`VowpalWabbit.Builder`	`readableModel(Path model)` Output human-readable final regressor with numeric features
`VowpalWabbit.Builder`	`recallTree(int k)` Use online tree for multiclass
`VowpalWabbit.Builder`	`redefine(String newNamespace, String... namespaces)` redefine namespaces beginning with characters of string S as namespace N.
`VowpalWabbit.Builder`	`replayB(String arg)` use experience replay at a specified level
`VowpalWabbit.Builder`	`replayC(String arg)` use experience replay at a specified level
`VowpalWabbit.Builder`	`replayM(String arg)` use experience replay at a specified level
`VowpalWabbit.Builder`	`ringSize(int ringSize)` size of example ring buffer
`VowpalWabbit.Builder`	`savePerPass()` Save the model after every pass over data
`VowpalWabbit.Builder`	`saveResume()` save extra state so learning can be resumed later with new data
`VowpalWabbit.Builder`	`search(int maxActionID)` Use learning to search, argument=maximum action id or 0 for LDF
`VowpalWabbit.Builder`	`sgd()` use regular stochastic gradient descent update.
`VowpalWabbit.Builder`	`skips(String namespace, int n)` Generate skips in N grams.
`VowpalWabbit.Builder`	`sortFeatures()` turn this on to disregard order in which features have been defined.
`VowpalWabbit.Builder`	`sparseL2(double l2)` use per feature normalized updates (=0)
`VowpalWabbit.Builder`	`spelling(String namespace)` compute spelling features for a give namespace (use '_' for default namespace)
`VowpalWabbit.Builder`	`stagePoly()` use stagewise polynomial feature learning
`VowpalWabbit.Builder`	`svrg()` Streaming Stochastic Variance Reduced Gradient
`VowpalWabbit.Builder`	`testonly()` Ignore label information and just test
`VowpalWabbit.Builder`	`top(int k)` top k recommendation
`VowpalWabbit.Builder`	`verbose()` Make vowpal wabbit writing debug and performance information to stderr
`VowpalWabbit.Builder`	`wapLdf(VowpalWabbit.LDF ldf)` Use weighted all-pairs multiclass learning with label dependent features.

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

- Method Detail
  - verbose
```
public VowpalWabbit.Builder verbose()
```
    Make vowpal wabbit writing debug and performance information to stderr
    
    Specified by:
    
    verbose in interface DebuggingOptions
    
    Returns:
    
    builder
  - randomSeed
```
public VowpalWabbit.Builder randomSeed(int seed)
```
    seed random number generator
    
    Specified by:
    
    randomSeed in interface MiscOptions
    
    Parameters:
    
    seed - random generator seed
    
    Returns:
    
    builder
  - ringSize
```
public VowpalWabbit.Builder ringSize(int ringSize)
```
    size of example ring buffer
    
    Parameters:
    
    ringSize - size of example ring
    
    Returns:
    
    builder
  - learningRate
```
public VowpalWabbit.Builder learningRate(double learningRate)
```
    Set learning rate
    
    Specified by:
    
    learningRate in interface UpdatesOptions
    
    Parameters:
    
    learningRate - learningRate learning rate. Must be positive
    
    Returns:
    
    builder
  - powerT
```
public VowpalWabbit.Builder powerT(double powerT)
```
    t power value
    
    Parameters:
    
    powerT - t power value
    
    Returns:
    
    builder
  - decayLearningRate
```
public VowpalWabbit.Builder decayLearningRate(double decay)
```
    Set Decay factor for learning_rate between passes
    
    Parameters:
    
    decay - exponential decay
    
    Returns:
    
    builder
  - initialT
```
public VowpalWabbit.Builder initialT(double initialT)
```
    initial t value
    
    Parameters:
    
    initialT - initial t value
    
    Returns:
    
    builder
  - featureMask
```
public VowpalWabbit.Builder featureMask(Path featureMask)
```
    Use existing regressor to determine which parameters may be updated.
    If no initialRegressor given, also used for initial weights.
    
    Specified by:
    
    featureMask in interface FeatureSelectionOptions
    
    Parameters:
    
    featureMask - path where to read feature mask
    
    Returns:
    
    builder
  - initialRegressor
```
public VowpalWabbit.Builder initialRegressor(Path initialRegressor)
```
    Initial regressor(s)
    
    Specified by:
    
    initialRegressor in interface OptionsToSaveAndLoadModel
    
    Parameters:
    
    initialRegressor - path where to read initial regressor
    
    Returns:
    
    builder
  - initialWeight
```
public VowpalWabbit.Builder initialWeight(double weight)
```
    Set all weights to an initial value of arg.
    
    Parameters:
    
    weight - initial weight value
    
    Returns:
    
    builder
  - randomWeights
```
public VowpalWabbit.Builder randomWeights(double arg)
```
    make initial weights random
    
    Parameters:
    
    arg - random maximum
    
    Returns:
    
    builder
  - inputFeatureRegularizer
```
public VowpalWabbit.Builder inputFeatureRegularizer(Path regularizationPath)
```
    Per feature regularization input file
    
    Parameters:
    
    regularizationPath - path to regularization input file
    
    Returns:
    
    builder
  - hash
```
public VowpalWabbit.Builder hash(VowpalWabbit.Hash hash)
```
    how to hash the features. Available options: strings, all
    
    Parameters:
    
    hash - hash strategy
    
    Returns:
    
    builder
  - ignore
```
public VowpalWabbit.Builder ignore(String namespace)
```
    ignore namespace <arg>
    
    Parameters:
    
    namespace - namespace name
    
    Returns:
    
    builder
  - keep
```
public VowpalWabbit.Builder keep(String namespace)
```
    keep namespace <arg>
    
    Parameters:
    
    namespace - namespace name
    
    Returns:
    
    builder
  - redefine
```
public VowpalWabbit.Builder redefine(String newNamespace,
                                     String... namespaces)
```
    redefine namespaces beginning with characters of string S as namespace N.
    <arg> shall be in form 'N:=S' where := is operator. Empty N or S are treated as default namespace.
    Use ':' as a wildcard in S.
    
    Parameters:
    
    newNamespace - new namespace name
    
    namespaces - old namespaces names
    
    Returns:
    
    builder
  - bitPrecision
```
public VowpalWabbit.Builder bitPrecision(int bitsNum)
```
    number of bits in the feature table.
    It mean feature table will have 2^bitsNum size
    
    Specified by:
    
    bitPrecision in interface OptionToExchangeRAMForQuality
    
    Parameters:
    
    bitsNum - bitsNum number of bits in hash feature table.
    
    Returns:
    
    builder
  - noconstant
```
public VowpalWabbit.Builder noconstant()
```
    Don't add a constant feature
    
    Specified by:
    
    noconstant in interface FeatureEngineeringFunctions
    
    Returns:
    
    builder
  - constant
```
public VowpalWabbit.Builder constant(double initialValue)
```
    Set initial value of constant
    
    Specified by:
    
    constant in interface FeatureEngineeringFunctions
    
    Parameters:
    
    initialValue - bias initial value
    
    Returns:
    
    builder
  - ngram
```
public VowpalWabbit.Builder ngram(String namespace,
                                  int n)
```
    Generate N grams. To generate N grams for a single namespace 'foo', arg should be fN.
    
    Specified by:
    
    ngram in interface FeatureEngineeringFunctions
    
    Parameters:
    
    namespace - namespace name, or ':' for any namespaces
    
    n - size of n-gram.
    
    Returns:
    
    builder
  - skips
```
public VowpalWabbit.Builder skips(String namespace,
                                  int n)
```
    Generate skips in N grams. This in conjunction with the ngram tag can be used to generate generalized n-skip-k-gram.
    To generate n-skips for a single namespace 'foo', arg should be fN.
    
    Specified by:
    
    skips in interface FeatureEngineeringFunctions
    
    Parameters:
    
    namespace - namespace name, or ':' for any namespaces
    
    n - size of skips n-gram.
    
    Returns:
    
    builder
  - featureLimit
```
public VowpalWabbit.Builder featureLimit(int n)
```
    limit to N features. To apply to a single namespace 'foo', arg should be fN
    
    Parameters:
    
    n - number of features
    
    Returns:
    
    builder
  - affix
```
public VowpalWabbit.Builder affix(String arg)
```
    generate prefixes/suffixes of features; argument '+2a,-3b,+1'
    means generate 2-char prefixes for namespace a, 3-char suffixes for b and 1 char
    prefixes for default namespace
    
    Parameters:
    
    arg - argument
    
    Returns:
    
    builder
  - spelling
```
public VowpalWabbit.Builder spelling(String namespace)
```
    compute spelling features for a give namespace (use '_' for default namespace)
    
    Parameters:
    
    namespace - namespace
    
    Returns:
    
    builder
  - dictionary
```
public VowpalWabbit.Builder dictionary(Path file)
```
    read a dictionary for additional features (arg either 'x:file' or just 'file')
    
    Parameters:
    
    file - dictionary path
    
    Returns:
    
    builder
  - dictionaryPath
```
public VowpalWabbit.Builder dictionaryPath(Path dir)
```
    look in this directory for dictionaries; defaults to current directory or env{PATH}
    
    Parameters:
    
    dir - dictionaries directory path
    
    Returns:
    
    builder
  - interactions
```
public VowpalWabbit.Builder interactions(String... namespaces)
```
    Create feature interactions of any level between namespaces.
    
    Parameters:
    
    namespaces - namspaces
    
    Returns:
    
    builder
  - permutations
```
public VowpalWabbit.Builder permutations()
```
    Use permutations instead of combinations for feature interactions of same namespace.
    
    Returns:
    
    builder
  - leaveDuplicateInteractions
```
public VowpalWabbit.Builder leaveDuplicateInteractions()
```
    Don't remove interactions with duplicate combinations of namespaces.
    For ex. this is a duplicate: '-q ab -q ba' and a lot more in '-q ::'.
    
    Returns:
    
    builder
  - quadratic
```
public VowpalWabbit.Builder quadratic(String firstNameSpace,
                                      String secondNamespace)
```
    Create and use quadratic features
    
    Specified by:
    
    quadratic in interface FeatureEngineeringFunctions
    
    Parameters:
    
    firstNameSpace - namespace or ":" for any
    
    secondNamespace - namespace or ":" for any
    
    Returns:
    
    builder
  - cubic
```
public VowpalWabbit.Builder cubic(String firstNameSpace,
                                  String secondNamespace,
                                  String thirdNamespace)
```
    Create and use cubic features
    
    Specified by:
    
    cubic in interface FeatureEngineeringFunctions
    
    Parameters:
    
    firstNameSpace - namespace or ":" for any
    
    secondNamespace - namespace or ":" for any
    
    thirdNamespace - namespace or ":" for any
    
    Returns:
    
    builder
  - testonly
```
public VowpalWabbit.Builder testonly()
```
    Ignore label information and just test
    
    Specified by:
    
    testonly in interface MiscOptions
    
    Returns:
    
    builder
  - holdoutPeriod
```
public VowpalWabbit.Builder holdoutPeriod(int holdout)
```
    holdout period for test only, default 10
    
    Parameters:
    
    holdout - holdout period size
    
    Returns:
    
    builder
  - holdoutAfter
```
public VowpalWabbit.Builder holdoutAfter(int n)
```
    holdout after n training examples, default off (disables holdoutPeriod)
    
    Parameters:
    
    n - number of examples in hodout
    
    Returns:
    
    builder
  - earlyTerminate
```
public VowpalWabbit.Builder earlyTerminate(int passes)
```
    Specify the number of passes tolerated when holdout loss doesn't decrease before early termination, default is 3
    
    Parameters:
    
    passes - number of passes
    
    Returns:
    
    builder
  - passes
```
public VowpalWabbit.Builder passes(int passes)
```
    Number of Training Passes
    
    Parameters:
    
    passes - number of passes
    
    Returns:
    
    builder
  - initialPassLength
```
public VowpalWabbit.Builder initialPassLength(int examples)
```
    initial number of examples per pass
    
    Parameters:
    
    examples - number of examples per pass
    
    Returns:
    
    builder
  - examples
```
public VowpalWabbit.Builder examples(int examples)
```
    number of examples to parse
    
    Parameters:
    
    examples - number of examples to parse
    
    Returns:
    
    builder
  - minPrediction
```
public VowpalWabbit.Builder minPrediction(double min)
```
    Smallest prediction to output
    
    Specified by:
    
    minPrediction in interface PredictionBoundaryOptions
    
    Parameters:
    
    min - minimum prediction, including
    
    Returns:
    
    builder
  - maxPrediction
```
public VowpalWabbit.Builder maxPrediction(double max)
```
    Largest prediction to output
    
    Specified by:
    
    maxPrediction in interface PredictionBoundaryOptions
    
    Parameters:
    
    max - maximum prediction, including
    
    Returns:
    
    builder
  - sortFeatures
```
public VowpalWabbit.Builder sortFeatures()
```
    turn this on to disregard order in which features have been defined. This will lead to smaller cache sizes
    
    Returns:
    
    builder
  - lossFunction
```
public VowpalWabbit.Builder lossFunction(Loss loss)
```
    Specify the loss function to be used, uses squared by default. Currently available ones are
    squared, classic, hinge, logistic, quantile and poisson. (=squared)
    
    Specified by:
    
    lossFunction in interface LinkAndLossOptions
    
    Parameters:
    
    loss - loss function
    
    Returns:
    
    builder
  - quantileTau
```
public VowpalWabbit.Builder quantileTau(double tau)
```
    Parameter \tau associated with Quantile loss. Defaults to 0.5 (=0.5)
    
    Specified by:
    
    quantileTau in interface LinkAndLossOptions
    
    Parameters:
    
    tau - tau parameter
    
    Returns:
    
    builder
  - l1
```
public VowpalWabbit.Builder l1(double l1)
```
    l_1 lambda
    
    Specified by:
    
    l1 in interface RegularizationOptions
    
    Parameters:
    
    l1 - l1 regularization. Must be not negative
    
    Returns:
    
    builder
  - l2
```
public VowpalWabbit.Builder l2(double l2)
```
    l_2 lambda
    
    Specified by:
    
    l2 in interface RegularizationOptions
    
    Parameters:
    
    l2 - l2 regularization. Must be not negative
    
    Returns:
    
    builder
  - namedLabels
```
public VowpalWabbit.Builder namedLabels(String... labels)
```
    use names for labels (multiclass, etc.) rather than integers, argument specified all possible labels, comma-sep,
    eg "--namedLabels Noun,Verb,Adj,Punc"
    
    Parameters:
    
    labels - labels
    
    Returns:
    
    builder
  - finalRegressor
```
public VowpalWabbit.Builder finalRegressor(Path regressor)
```
    Final regressor
    
    Specified by:
    
    finalRegressor in interface OptionsToSaveAndLoadModel
    
    Parameters:
    
    regressor - path where to store final regressor
    
    Returns:
    
    builder
  - readableModel
```
public VowpalWabbit.Builder readableModel(Path model)
```
    Output human-readable final regressor with numeric features
    
    Specified by:
    
    readableModel in interface DebuggingOptions
    
    Parameters:
    
    model - path where to store readable model
    
    Returns:
    
    builder
  - saveResume
```
public VowpalWabbit.Builder saveResume()
```
    save extra state so learning can be resumed later with new data
    
    Returns:
    
    builder
  - savePerPass
```
public VowpalWabbit.Builder savePerPass()
```
    Save the model after every pass over data
    
    Returns:
    
    builder
  - outputFeatureRegularizerBinary
```
public VowpalWabbit.Builder outputFeatureRegularizerBinary(Path regularizationFile)
```
    Per feature regularization output file
    
    Parameters:
    
    regularizationFile - path where to store regularization output file
    
    Returns:
    
    builder
  - outputFeatureRegularizerText
```
public VowpalWabbit.Builder outputFeatureRegularizerText(Path regularizationFile)
```
    Per feature regularization output file, in text
    
    Parameters:
    
    regularizationFile - path where to store regularization output file
    
    Returns:
    
    builder
  - id
```
public VowpalWabbit.Builder id(String id)
```
    User supplied ID embedded into the final regressor
    
    Parameters:
    
    id - model id
    
    Returns:
    
    builder
  - auditRegressor
```
public VowpalWabbit.Builder auditRegressor(Path regressor)
```
    stores feature names and their regressor values.
    Same dataset must be used for both regressor training and this mode.
    
    Parameters:
    
    regressor - path where to read regressor for audit
    
    Returns:
    
    builder
  - bootstrap
```
public VowpalWabbit.Builder bootstrap(int k)
```
    k-way bootstrap by online importance resampling
    
    Parameters:
    
    k - number of bootstrap resamples
    
    Returns:
    
    builder
  - search
```
public VowpalWabbit.Builder search(int maxActionID)
```
    Use learning to search, argument=maximum action id or 0 for LDF
    
    Parameters:
    
    maxActionID - max action id
    
    Returns:
    
    builder
  - replayC
```
public VowpalWabbit.Builder replayC(String arg)
```
    use experience replay at a specified level
    [b=classification/regression, m=multiclass, c=cost sensitive] with specified buffer size
    
    Parameters:
    
    arg - argument
    
    Returns:
    
    builder
  - cbify
```
public VowpalWabbit.Builder cbify(int k)
```
    Convert multiclass on <k> classes into a contextual bandit problem
    
    Parameters:
    
    k - number of classes
    
    Returns:
    
    builder
  - cbExploreAdf
```
public VowpalWabbit.Builder cbExploreAdf()
```
    Online explore-exploit for a contextual bandit problem with multiline action dependent features
    
    Returns:
    
    builder
  - cbExplore
```
public VowpalWabbit.Builder cbExplore(int k)
```
    Online explore-exploit for a <k> action contextual bandit problem
    
    Parameters:
    
    k - number of actions
    
    Returns:
    
    builder
  - multiworldTest
```
public VowpalWabbit.Builder multiworldTest(String arg)
```
    Evaluate features as a policies
    
    Parameters:
    
    arg - argument
    
    Returns:
    
    builder
  - cbAdf
```
public VowpalWabbit.Builder cbAdf()
```
    Do Contextual Bandit learning with multiline action dependent features.
    
    Returns:
    
    builder
  - cb
```
public VowpalWabbit.Builder cb(int k)
```
    Use contextual bandit learning with <k> costs
    
    Parameters:
    
    k - number of costs
    
    Returns:
    
    builder
  - csoaaLdf
```
public VowpalWabbit.Builder csoaaLdf(VowpalWabbit.LDF ldf)
```
    Use one-against-all multiclass learning with label dependent features. Specify singleline or multiline.
    
    Parameters:
    
    ldf - ldf
    
    Returns:
    
    builder
  - wapLdf
```
public VowpalWabbit.Builder wapLdf(VowpalWabbit.LDF ldf)
```
    Use weighted all-pairs multiclass learning with label dependent features. Specify singleline or multiline.
    
    Parameters:
    
    ldf - ldf
    
    Returns:
    
    builder
  - interact
```
public VowpalWabbit.Builder interact(String arg)
```
    Put weights on feature products from namespaces <n1> and <n2>
    
    Parameters:
    
    arg - argument
    
    Returns:
    
    builder
  - csoaa
```
public VowpalWabbit.Builder csoaa(int k)
```
    One-against-all multiclass with <k> costs
    
    Parameters:
    
    k - number of costs
    
    Returns:
    
    builder
  - multilabelOaa
```
public VowpalWabbit.Builder multilabelOaa(int k)
```
    One-against-all multilabel with <k> labels
    
    Parameters:
    
    k - number of labels
    
    Returns:
    
    builder
  - recallTree
```
public VowpalWabbit.Builder recallTree(int k)
```
    Use online tree for multiclass
    
    Parameters:
    
    k - number of classes
    
    Returns:
    
    builder
  - logMulti
```
public VowpalWabbit.Builder logMulti(int k)
```
    Use online tree for multiclass
    
    Parameters:
    
    k - number of classes
    
    Returns:
    
    builder
  - ect
```
public VowpalWabbit.Builder ect(int k)
```
    Error correcting tournament with <k> labels
    
    Parameters:
    
    k - number of labels
    
    Returns:
    
    builder
  - boosting
```
public VowpalWabbit.Builder boosting(int n)
```
    Online boosting with <N> weak learners
    
    Parameters:
    
    n - number of weak learners
    
    Returns:
    
    builder
  - oaa
```
public VowpalWabbit.Builder oaa(int k)
```
    One-against-all multiclass with <k> labels
    
    Parameters:
    
    k - number of classes
    
    Returns:
    
    builder
  - top
```
public VowpalWabbit.Builder top(int k)
```
    top k recommendation
    
    Parameters:
    
    k - number of top recomendations
    
    Returns:
    
    builder
  - replayM
```
public VowpalWabbit.Builder replayM(String arg)
```
    use experience replay at a specified level
    [b=classification/regression, m=multiclass, c=cost sensitive] with specified buffer size
    
    Parameters:
    
    arg - argument
    
    Returns:
    
    builder
  - binary
```
public VowpalWabbit.Builder binary()
```
    report loss as binary classification on -1,1
    
    Returns:
    
    builder
  - link
```
public VowpalWabbit.Builder link(Link link)
```
    Specify the link function: identity, logistic, glf1 or poisson (=identity)
    
    Specified by:
    
    link in interface LinkAndLossOptions
    
    Parameters:
    
    link - link function
    
    Returns:
    
    builder
  - stagePoly
```
public VowpalWabbit.Builder stagePoly()
```
    use stagewise polynomial feature learning
    
    Returns:
    
    builder
  - lrqfa
```
public VowpalWabbit.Builder lrqfa(String firstNamespace,
                                  String secondNamespace,
                                  int k)
```
    use low rank quadratic features with field aware weights
    
    Specified by:
    
    lrqfa in interface FeatureEngineeringFunctions
    
    Parameters:
    
    firstNamespace - - namespace or ":" for any
    
    secondNamespace - - namespace or ":" for any
    
    k - factorized matrices width
    
    Returns:
    
    builder
  - lrq
```
public VowpalWabbit.Builder lrq(String firstNamespace,
                                String secondNamespace,
                                int k)
```
    use low rank quadratic features
    
    Parameters:
    
    firstNamespace - - namespace or ":" for any
    
    secondNamespace - - namespace or ":" for any
    
    k - factorized matrices width
    
    Returns:
    
    builder
  - lrqdropout
```
public VowpalWabbit.Builder lrqdropout()
```
    use dropout training for low rank quadratic features
    
    Returns:
    
    builder
  - autolink
```
public VowpalWabbit.Builder autolink(int d)
```
    create link function with polynomial d
    
    Parameters:
    
    d - polynomial degree
    
    Returns:
    
    builder
  - newMf
```
public VowpalWabbit.Builder newMf(int rank)
```
    rank for reduction-based matrix factorization
    
    Parameters:
    
    rank - rank
    
    Returns:
    
    builder
  - nn
```
public VowpalWabbit.Builder nn(int units)
```
    Sigmoidal feedforward network with <k> hidden units
    
    Parameters:
    
    units - number of hidden units
    
    Returns:
    
    builder
  - confidenceAfterTraining
```
public VowpalWabbit.Builder confidenceAfterTraining()
```
    Confidence after training
    
    Returns:
    
    builder
  - confidence
```
public VowpalWabbit.Builder confidence()
```
    Get confidence for binary predictions
    
    Returns:
    
    builder
  - activeCover
```
public VowpalWabbit.Builder activeCover()
```
    enable active learning with cover
    
    Returns:
    
    builder
  - active
```
public VowpalWabbit.Builder active()
```
    enable active learning
    
    Returns:
    
    builder
  - replayB
```
public VowpalWabbit.Builder replayB(String arg)
```
    use experience replay at a specified level
    [b=classification/regression, m=multiclass, c=cost sensitive] with specified buffer size
    
    Parameters:
    
    arg - argument
    
    Returns:
    
    builder
  - ojaNewton
```
public VowpalWabbit.Builder ojaNewton()
```
    Online Newton with Oja's Sketch
    
    Returns:
    
    builder
  - bfgs
```
public VowpalWabbit.Builder bfgs()
```
    use bfgs optimization
    
    Returns:
    
    builder
  - conjugateGradient
```
public VowpalWabbit.Builder conjugateGradient()
```
    use conjugate gradient based optimization
    
    Returns:
    
    builder
  - lda
```
public VowpalWabbit.Builder lda(int topics)
```
    Run lda with <int> topics
    
    Parameters:
    
    topics - number of lda topics
    
    Returns:
    
    builder
  - noop
```
public VowpalWabbit.Builder noop()
```
    do no learning
    
    Returns:
    
    builder
  - rank
```
public VowpalWabbit.Builder rank(int rank)
```
    rank for matrix factorization.
    
    Parameters:
    
    rank - rank for matrix factorization
    
    Returns:
    
    builder
  - svrg
```
public VowpalWabbit.Builder svrg()
```
    Streaming Stochastic Variance Reduced Gradient
    
    Returns:
    
    builder
  - ftrl
```
public VowpalWabbit.Builder ftrl()
```
    FTRL: Follow the Proximal Regularized Leader
    
    Specified by:
    
    ftrl in interface FeatureSelectionOptions
    
    Returns:
    
    builder
  - pistol
```
public VowpalWabbit.Builder pistol()
```
    FTRL: Parameter-free Stochastic Learning
    
    Returns:
    
    builder
  - ksvm
```
public VowpalWabbit.Builder ksvm()
```
    kernel svm
    
    Returns:
    
    builder
  - sgd
```
public VowpalWabbit.Builder sgd()
```
    use regular stochastic gradient descent update.
    
    Specified by:
    
    sgd in interface UpdatesOptions
    
    Returns:
    
    builder
  - adaptive
```
public VowpalWabbit.Builder adaptive()
```
    use adaptive, individual learning rates.
    
    Specified by:
    
    adaptive in interface UpdatesOptions
    
    Returns:
    
    builder
  - invariant
```
public VowpalWabbit.Builder invariant()
```
    use safe/importance aware updates.
    
    Specified by:
    
    invariant in interface UpdatesOptions
    
    Returns:
    
    builder
  - normalized
```
public VowpalWabbit.Builder normalized()
```
    use per feature normalized updates
    
    Specified by:
    
    normalized in interface UpdatesOptions
    
    Returns:
    
    builder
  - sparseL2
```
public VowpalWabbit.Builder sparseL2(double l2)
```
    use per feature normalized updates (=0)
    
    Parameters:
    
    l2 - l2 regularization. Must be not negative
    
    Returns:
    
    builder
  - cache
```
public VowpalWabbit.Builder cache()
```
    Use a cache. The default is <data>.cache
    
    Returns:
    
    builder
  - cacheFile
```
public VowpalWabbit.Builder cacheFile(Path cacheFile)
```
    The location(s) of cacheFile.
    
    Parameters:
    
    cacheFile - path to cache file
    
    Returns:
    
    builder
  - killCache
```
public VowpalWabbit.Builder killCache()
```
    do not reuse existing cache: create a new one always
    
    Returns:
    
    builder
  - compressed
```
public VowpalWabbit.Builder compressed()
```
    use gzip format whenever possible. If a cache file is being created,
    this option creates a compressed cache file.
    A mixture of raw-text and compressed inputs are supported with autodetection.
    
    Returns:
    
    builder
  - parameter
```
public VowpalWabbit.Builder parameter(String argumentLine)
```
    Add vowpal wabit argument
    
    Parameters:
    
    argumentLine - parameter line
    
    Returns:
    
    builder
  - getCommand
```
public String getCommand()
```
    Get command option will be passes to VWLearner
    
    Returns:
    
    command options
  - buildFloatLearner
```
public com.indeed.vw.wrapper.learner.VWFloatLearner buildFloatLearner()
```
    Description copied from interface: SGDVowpalWabbitBuilder
    
    Build learner
    
    Specified by:
    
    buildFloatLearner in interface SGDVowpalWabbitBuilder
    
    Returns:
    
    VWIntLearner object
  - buildIntLearner
```
public com.indeed.vw.wrapper.learner.VWIntLearner buildIntLearner()
```
    Returns:
    
    VWIntLearner object
  - buildFloatArrayLearner
```
public com.indeed.vw.wrapper.learner.VWFloatArrayLearner buildFloatArrayLearner()
```
    Returns:
    
    VWFloatArrayLearner object
  - buildIntArrayLearner
```
public com.indeed.vw.wrapper.learner.VWIntArrayLearner buildIntArrayLearner()
```
    Returns:
    
    VWIntArrayLearner object

Class VowpalWabbit.Builder

Method Summary

Methods inherited from class java.lang.Object

Method Detail

verbose

randomSeed

ringSize

learningRate

powerT

decayLearningRate

initialT

featureMask

initialRegressor

initialWeight

randomWeights

inputFeatureRegularizer

hash

ignore

keep

redefine

bitPrecision

noconstant

constant

ngram

skips

featureLimit

affix

spelling

dictionary

dictionaryPath

interactions

permutations

leaveDuplicateInteractions

quadratic

cubic

testonly

holdoutPeriod

holdoutAfter

earlyTerminate

passes

initialPassLength

examples

minPrediction

maxPrediction

sortFeatures

lossFunction

quantileTau

l1

l2

namedLabels

finalRegressor

readableModel

saveResume

savePerPass

outputFeatureRegularizerBinary

outputFeatureRegularizerText

id

auditRegressor

bootstrap

search

replayC

cbify

cbExploreAdf

cbExplore

multiworldTest

cbAdf

cb

csoaaLdf

wapLdf

interact

csoaa

multilabelOaa

recallTree

logMulti

ect

boosting

oaa

top

replayM

binary