feature

Type Members

1. case class BucketizerModel(splits: Array[Double]) extends Serializable with Product

   

   Class for a bucketizer model.

   Class for a bucketizer model.

   Bucketizer will place incoming feature into a bucket.

   splits

   splits used to determine bucket

   Annotations

   @SparkCode()

2. case class ElementwiseProductModel(scalingVec: Vector) extends Serializable with Product

   

   Class for an element wise product model.

   Class for an element wise product model.

   scalingVec

   vector for scaling feature vectors

   Annotations

   @SparkCode()

3. case class HashingTermFrequencyModel(numFeatures: Int = 1 << 18, binary: Boolean = false) extends Product with Serializable

   

   Class for hashing token frequencies into a vector.

   Class for hashing token frequencies into a vector.

   Source adapted from: Apache Spark Utils and HashingTF, see NOTICE for contributors

   numFeatures

   size of feature vector to hash into

   Annotations

   @SparkCode()

4. case class MaxAbsScalerModel(maxAbs: Vector) extends Serializable with Product

   

   Class for MaxAbs Scaler model.

   Class for MaxAbs Scaler model.

   maxAbs

   max absolute value

   Annotations

   @SparkCode()

5. case class MinMaxScalerModel(originalMin: Vector, originalMax: Vector) extends Serializable with Product

   

   Class for MinMax Scaler Transformer

   Class for MinMax Scaler Transformer

   MinMax Scaler will use the Min/Max values to scale input features.

   originalMin

   minimum values from training features

   originalMax

   maximum values from training features

   Annotations

   @SparkCode()

6. case class NGramModel(n: Int) extends Serializable with Product

   

   Created by mikhail on 9/29/16.

7. case class NormalizerModel(pNorm: Double) extends Serializable with Product

   

   Class for storing a normalizer model.

   Class for storing a normalizer model.

   pNorm

   p normalization param

   Annotations

   @SparkCode()

8. case class OneHotEncoderModel(size: Int) extends Serializable with Product

   

   Class for a one hot encoder model.

   Class for a one hot encoder model.

   One hot encoders are used to vectorize nominal features in preparation for models such as linear regression or logistic regression where binary and not multinomial features are supported in the feature vector.

   size

   size of the output one hot vectors

9. case class PcaModel(principalComponents: DenseMatrix) extends Product with Serializable

   

   Class for principal components analysis model.

   Class for principal components analysis model.

   principalComponents

   matrix of principal components

10. case class PolynomialExpansionModel(degree: Int) extends Serializable with Product

    

    Created by mikhail on 10/16/16.

    Created by mikhail on 10/16/16.

    Annotations

    @SparkCode()

11. case class ReverseStringIndexerModel(labels: Seq[String]) extends Product with Serializable

    

    Class for a reverse string indexer model.

    Class for a reverse string indexer model.

    This model reverses the StringIndexerModel model. Use this to go from an integer representation of a label to a string.

    labels

    labels for reverse string indexing

12. case class StandardScalerModel(std: Option[Vector], mean: Option[Vector]) extends Serializable with Product

    

    Class for standard scaler models.

    Class for standard scaler models.

    Standard scaler will use stddev, mean, or both to scale a feature vector down.

    std

    optional standard deviations of features

    mean

    optional means of features

    Annotations

    @SparkCode()

13. case class StopWordsRemoverModel(stopWords: Array[String], caseSensitive: Boolean) extends Serializable with Product

    

    Created by mikhail on 10/16/16.

14. case class StringIndexerModel(labels: Seq[String]) extends Serializable with Product

    

    Class for string indexer model.

    Class for string indexer model.

    String indexer converts a string into an integer representation.

    labels

    list of labels that can be indexed

15. case class TokenizerModel(regex: String = "\\s") extends Product with Serializable

    

    Class for a tokenizer model.

    Class for a tokenizer model.

    Default regular expression for tokenizing strings is defined by TokenizerModel.defaultTokenizer

    regex

    regular expression used for tokenizing strings

16. case class VectorAssemblerModel() extends Serializable with Product

    

    Class for a vector assembler model.

    Class for a vector assembler model.

    Vector assemblers take an input set of doubles and vectors and create a new vector out of them. This is primarily used to get all desired features into one vector before training a model.

    Annotations

    @SparkCode()