Package ai.djl.training.evaluator

Class Coverage

java.lang.Object
ai.djl.training.evaluator.Evaluator
ai.djl.training.evaluator.AbstractAccuracy
ai.djl.training.evaluator.Coverage
public class Coverage extends AbstractAccuracy
Coverage for a Regression problem: it measures the percent of predictions greater than the actual target, to determine whether the predictor is over-forecasting or under-forecasting. e.g. 0.50 if we predict near the median of the distribution. 
  def coverage(target, forecast):
     return (np.mean((target < forecast)))
...

  • Constructor Details

    • Coverage

      public Coverage()
      Creates an evaluator that measures the percent of predictions greater than the actual target.

    • Coverage

      public Coverage(String name, int axis)
      Creates an evaluator that measures the percent of predictions greater than the actual target.
      Parameters:
      name - the name of the evaluator, default is "Coverage"
      axis - the axis along which to count the correct prediction, default is 1

  • Method Details

    • accuracyHelper

      protected ai.djl.util.Pair<Long,NDArray> accuracyHelper(NDList labels, NDList predictions)
      A helper for classes extending AbstractAccuracy.
      Specified by:
      accuracyHelper in class AbstractAccuracy
      Parameters:
      labels - the labels to get accuracy for
      predictions - the predictions to get accuracy for
      Returns:
      a pair(number of total values, ndarray int of correct values)