public OvaPipelineItem(ITrainerInputWithLabel trainer, bool useProbabilities)
{
_trainer = trainer;
_useProbabilities = useProbabilities;
}
/// <summary>
/// One-versus-all, OvA, learner (also known as One-vs.-rest, "OvR") is a multi-class learner
/// with the strategy to fit one binary classifier per class in the dataset.
/// It trains the provided binary classifier for each class against the other classes, where the current
/// class is treated as the positive labels and examples in other classes are treated as the negative classes.
/// See <a href="https://en.wikipedia.org/wiki/Multiclass_classification#One-vs.-rest">wikipedia</a> page.
/// </summary>
///<example>
/// In order to use it all you need to do is add it to pipeline as regular learner:
///
/// pipeline.Add(OneVersusAll.With(new StochasticDualCoordinateAscentBinaryClassifier()));
/// </example>
/// <remarks>
/// The base trainer must be a binary classifier. To check the available binary classifiers, type BinaryClassifiers,
/// and look at the available binary learners as suggested by IntelliSense.
/// </remarks>
/// <param name="trainer">Underlying binary trainer</param>
/// <param name="useProbabilities">"Use probabilities (vs. raw outputs) to identify top-score category.
/// By specifying it to false, you can tell One-versus-all to not use the probabilities but instead
/// the raw uncalibrated scores from each predictor. This is generally not recommended, since these quantities
/// are not meant to be comparable from one predictor to another, unlike calibrated probabilities.</param>
public static ILearningPipelineItem With(ITrainerInputWithLabel trainer, bool useProbabilities = true)
{
return(new OvaPipelineItem(trainer, useProbabilities));
}
