Package smile.validation

Class ClassificationValidation<M>

java.lang.Object
smile.validation.ClassificationValidation<M>
Type Parameters:
M - the model type.
All Implemented Interfaces:
Serializable
public class ClassificationValidation<M> extends Object implements Serializable
Classification model validation results.
See Also:

  • Field Details

    • model

      public final M model
      The model.

    • truth

      public final int[] truth
      The true class labels of validation data.

    • prediction

      public final int[] prediction
      The model prediction.

    • posteriori

      public final double[][] posteriori
      The posteriori probability of prediction if the model is a soft classifier.

    • confusion

      public final ConfusionMatrix confusion
      The confusion matrix.

    • metrics

      public final ClassificationMetrics metrics
      The classification metrics.

  • Constructor Details

    • ClassificationValidation

      public ClassificationValidation(M model, double fitTime, double scoreTime, int[] truth, int[] prediction)
      Constructor.
      Parameters:
      model - the model.
      fitTime - the time in milliseconds of fitting the model.
      scoreTime - the time in milliseconds of scoring the validation data.
      truth - the ground truth.
      prediction - the predictions.

    • ClassificationValidation

      public ClassificationValidation(M model, double fitTime, double scoreTime, int[] truth, int[] prediction, double[][] posteriori)
      Constructor of soft classifier validation.
      Parameters:
      model - the model.
      fitTime - the time in milliseconds of fitting the model.
      scoreTime - the time in milliseconds of scoring the validation data.
      truth - the ground truth.
      prediction - the predictions.
      posteriori - the posteriori probabilities of predictions.

  • Method Details

    • toString

      public String toString()
      Overrides:
      toString in class Object

    • of

      public static <T, M extends Classifier<T>> ClassificationValidation<M> of(T[] x, int[] y, T[] testx, int[] testy, BiFunction<T[],int[],M> trainer)
      Trains and validates a model on a train/validation split.
      Type Parameters:
      T - the data type of samples.
      M - the model type.
      Parameters:
      x - the training data.
      y - the class labels of training data.
      testx - the validation data.
      testy - the class labels of validation data.
      trainer - the lambda to train the model.
      Returns:
      the validation results.

    • of

      public static <T, M extends Classifier<T>> ClassificationValidations<M> of(Bag[] bags, T[] x, int[] y, BiFunction<T[],int[],M> trainer)
      Trains and validates a model on multiple train/validation split.
      Type Parameters:
      T - the data type of samples.
      M - the model type.
      Parameters:
      bags - the data splits.
      x - the training data.
      y - the class labels.
      trainer - the lambda to train the model.
      Returns:
      the validation results.

    • of

      public static <M extends DataFrameClassifier> ClassificationValidation<M> of(smile.data.formula.Formula formula, smile.data.DataFrame train, smile.data.DataFrame test, BiFunction<smile.data.formula.Formula,smile.data.DataFrame,M> trainer)
      Trains and validates a model on a train/validation split.
      Type Parameters:
      M - the model type.
      Parameters:
      formula - the model formula.
      train - the training data.
      test - the validation data.
      trainer - the lambda to train the model.
      Returns:
      the validation results.

    • of

      public static <M extends DataFrameClassifier> ClassificationValidations<M> of(Bag[] bags, smile.data.formula.Formula formula, smile.data.DataFrame data, BiFunction<smile.data.formula.Formula,smile.data.DataFrame,M> trainer)
      Trains and validates a model on multiple train/validation split.
      Type Parameters:
      M - the model type.
      Parameters:
      bags - the data splits.
      formula - the model formula.
      data - the data.
      trainer - the lambda to train the model.
      Returns:
      the validation results.