Type Members

1.  case class FeatureRotator(baseLearner: Learner) extends Learner with Product with Serializable

   Rotate the training data before passing along to a base learner

   Rotate the training data before passing along to a base learner

   This may be useful for improving randomization in random forests, especially when using random feature selection without bagging.

   Created by gregor-robinson on 2020-01-02.

2.  case class MultiTaskFeatureRotator(baseLearner: MultiTaskLearner) extends MultiTaskLearner with Product with Serializable
3.  case class MultiTaskRotatedFeatureTrainingResult(baseTrainingResult: MultiTaskTrainingResult, rotatedFeatures: IndexedSeq[Int], trans: DenseMatrix[Double]) extends MultiTaskTrainingResult with Product with Serializable

   Training result bundling the base learner's multitask training result with the list of rotated features and the transformation

   Training result bundling the base learner's multitask training result with the list of rotated features and the transformation

   baseTrainingResult

   training result to which to delegate prediction on rotated features

   rotatedFeatures

   indices of features to rotate

   trans

   matrix to apply to features

4.  class MultiTaskStandardizer extends MultiTaskLearner
5.  class MultiTaskStandardizerTrainingResult extends MultiTaskTrainingResult

   Training result wrapping the base learner's training result next to the transformations

6.  case class RotatedFeatureModel[T](baseModel: Model[PredictionResult[T]], rotatedFeatures: IndexedSeq[Int], trans: DenseMatrix[Double]) extends Model[PredictionResult[T]] with Product with Serializable

   Model bundling the base learner's model with the list of rotated features and the transformation

   Model bundling the base learner's model with the list of rotated features and the transformation

   T

   label type

   baseModel

   model to which to delegate prediction on rotated features

   rotatedFeatures

   indices of features to rotate

   trans

   matrix to apply to features

7.  case class RotatedFeaturePrediction[T](baseResult: PredictionResult[T], rotatedFeatures: IndexedSeq[Int], trans: DenseMatrix[Double]) extends PredictionResult[T] with Product with Serializable

   Prediction bundling the base learner's prediction with the list of rotated features and the transformation

   Prediction bundling the base learner's prediction with the list of rotated features and the transformation

   T

   label type

8.  case class RotatedFeatureTrainingResult(baseTrainingResult: TrainingResult, rotatedFeatures: IndexedSeq[Int], trans: DenseMatrix[Double]) extends TrainingResult with Product with Serializable

   Training result bundling the base learner's training result with the list of rotated features and the transformation

   Training result bundling the base learner's training result with the list of rotated features and the transformation

   baseTrainingResult

   training result to which to delegate prediction on rotated features

   rotatedFeatures

   indices of features to rotate

   trans

   matrix to apply to features

9.  case class Standardization(shift: Double, scale: Double) extends Product with Serializable
10.  case class Standardizer(baseLearner: Learner) extends Learner with Product with Serializable

    Standardize the training data to zero mean and unit variance before feeding it into another learner

    Standardize the training data to zero mean and unit variance before feeding it into another learner

    This is particularly helpful for regularized methods, like ridge regression, where the relative scale of the features and labels is important.

    Created by maxhutch on 2/19/17.

11.  class StandardizerModel[T] extends Model[PredictionResult[T]]

    Model that wraps the base model next to the transformations

    Model that wraps the base model next to the transformations

    T

    type of prediction

12.  class StandardizerPrediction[T] extends PredictionResult[T]

    Prediction that wraps the base prediction next to the transformation

    Prediction that wraps the base prediction next to the transformation

    T

    type of prediction

13.  class StandardizerTrainingResult extends TrainingResult

    Training result wrapping the base learner's training result next to the transformations