Package com.tibco.patterns.learn.api.project

Class DatasetResult

  • public final class DatasetResult
extends java.lang.Object
    Stores model training results for one dataset. Objects of DatasetResult are returned from Learn API. No need to construct them in Learn UI code. It is a read-only view of COErrorRate.IterationResult that provides access to specific statistics used in Learn UI. It could potentially provide a view to a result class of another convergence observer. Client code may construct DatasetResult independently from IterationResult. It may also obtain a copy of the IterationResult object to get access to the advanced statistics that do not have getters in DatasetResult.

    • Constructor Detail

      • DatasetResult

        public DatasetResult​(COErrorRate.IterationResult coResult)
        Copies information from given IterationResult to this object. Can be used to avoid creating DatasetResultType when dataset results are not stored in XML.
        Parameters:
        coResult - - result from model training or evaluation. Not null.

    • Method Detail

      • toString

        public java.lang.String toString()
        Overrides:
        toString in class java.lang.Object

      • getIterationResult

        public COErrorRate.IterationResult getIterationResult()
        May be used to get access to statistics not directly exposed by DatasetResult methods. Modifying the returned object has no effect on this DatasetResult object.
        Returns:
        a copy of the IterationResult object stored in DatasetResult.

      • errorRate

        public double errorRate()
        Returns:
        model error rate

      • falsePosRate

        public double falsePosRate()
        Returns:
        false positive rate

      • falseNegRate

        public double falseNegRate()
        Returns:
        false negative rate

      • noPredictionRate

        public double noPredictionRate()
        Returns:
        the rate of examples that can't be predicted by this model

      • meanConf

        public double meanConf()
        Returns:
        mean prediction confidence (for all examples)

      • isConverged

        public boolean isConverged()
        Use this method with validation dataset result.
        Returns:
        whether the error rate is below the COErrorRate's default target error rate.

      • isPerfectResult

        public boolean isPerfectResult()
        Returns:
        true if all examples in the dataset were predicted correctly.