Package org.apache.spark.mllib.evaluation

Class MulticlassMetrics

Object
org.apache.spark.mllib.evaluation.MulticlassMetrics
public class MulticlassMetrics extends Object
Evaluator for multiclass classification.

param: predictionAndLabels an RDD of (prediction, label, weight, probability) or (prediction, label, weight) or (prediction, label) tuples.

  • Constructor Details

    • MulticlassMetrics

      public MulticlassMetrics(RDD<? extends scala.Product> predictionAndLabels)

  • Method Details

    • accuracy

      public double accuracy()

    • confusionMatrix

      public Matrix confusionMatrix()
      Returns confusion matrix: predicted classes are in columns, they are ordered by class label ascending, as in "labels"
      Returns:
      (undocumented)

    • fMeasure

      public double fMeasure(double label, double beta)
      Returns f-measure for a given label (category)
      Parameters:
      label - the label.
      beta - the beta parameter.
      Returns:
      (undocumented)

    • fMeasure

      public double fMeasure(double label)
      Returns f1-measure for a given label (category)
      Parameters:
      label - the label.
      Returns:
      (undocumented)

    • falsePositiveRate

      public double falsePositiveRate(double label)
      Returns false positive rate for a given label (category)
      Parameters:
      label - the label.
      Returns:
      (undocumented)

    • hammingLoss

      public double hammingLoss()

    • labels

      public double[] labels()

    • logLoss

      public double logLoss(double eps)
      Returns the log-loss, aka logistic loss or cross-entropy loss.
      Parameters:
      eps - log-loss is undefined for p=0 or p=1, so probabilities are clipped to max(eps, min(1 - eps, p)).
      Returns:
      (undocumented)

    • precision

      public double precision(double label)
      Returns precision for a given label (category)
      Parameters:
      label - the label.
      Returns:
      (undocumented)

    • recall

      public double recall(double label)
      Returns recall for a given label (category)
      Parameters:
      label - the label.
      Returns:
      (undocumented)

    • truePositiveRate

      public double truePositiveRate(double label)
      Returns true positive rate for a given label (category)
      Parameters:
      label - the label.
      Returns:
      (undocumented)

    • weightedFMeasure

      public double weightedFMeasure(double beta)
      Returns weighted averaged f-measure
      Parameters:
      beta - the beta parameter.
      Returns:
      (undocumented)

    • weightedFMeasure

      public double weightedFMeasure()

    • weightedFalsePositiveRate

      public double weightedFalsePositiveRate()

    • weightedPrecision

      public double weightedPrecision()

    • weightedRecall

      public double weightedRecall()

    • weightedTruePositiveRate

      public double weightedTruePositiveRate()