Class EpisodeRunner.EpisodeResult

java.lang.Object
neqsim.process.ml.EpisodeRunner.EpisodeResult
All Implemented Interfaces:
Serializable
Enclosing class:
EpisodeRunner

public static class EpisodeRunner.EpisodeResult extends Object implements Serializable
Result of running an episode.
Version:
1.0
Author:
ESOL
See Also:
  • Field Details

    • serialVersionUID

      private static final long serialVersionUID
      See Also:
    • totalReward

      public final double totalReward
      Total reward accumulated.
    • steps

      public final int steps
      Number of steps taken.
    • terminated

      public final boolean terminated
      Whether episode terminated (vs truncated).
    • observations

      public final List<double[]> observations
      History of observations.
    • actions

      public final List<double[]> actions
      History of actions.
    • rewards

      public final List<Double> rewards
      History of rewards.
    • finalObservation

      public final double[] finalObservation
      Final observation.
  • Constructor Details

    • EpisodeResult

      public EpisodeResult(double totalReward, int steps, boolean terminated, List<double[]> observations, List<double[]> actions, List<Double> rewards, double[] finalObservation)
      Constructor.
      Parameters:
      totalReward - total reward
      steps - steps taken
      terminated - episode terminated
      observations - observation history
      actions - action history
      rewards - reward history
      finalObservation - final observation
  • Method Details

    • getMeanReward

      public double getMeanReward()
      Get mean reward per step.
      Returns:
      mean reward
    • getObservation

      public double getObservation(int index, int featureIndex)
      Get observation at specific index.
      Parameters:
      index - observation index
      featureIndex - feature index within observation
      Returns:
      feature value
    • getFeatureTrajectory

      public double[] getFeatureTrajectory(int featureIndex)
      Get feature trajectory as array.
      Parameters:
      featureIndex - feature index
      Returns:
      array of feature values over time