/**
* As regression data set dataset.
*
* @param featurizer the featurizer
* @param labelAttributeType the label attribute
* @return the dataset
*/
default Dataset<Instance> asRegressionDataSet(
@NonNull Featurizer<HString> featurizer, @NonNull AttributeType labelAttributeType) {
return Dataset.regression()
.type(getDataSetType())
.source(asLabeledStream(labelAttributeType).map(featurizer::extractLabeled))
.build();
}
/**
* As regression data set dataset.
*
* @param featurizer the featurizer
* @param labelFunction the label function
* @return the dataset
*/
default Dataset<Instance> asRegressionDataSet(
@NonNull Featurizer<HString> featurizer,
@NonNull SerializableFunction<HString, Double> labelFunction) {
return Dataset.regression()
.type(getDataSetType())
.source(asLabeledStream(labelFunction).map(featurizer::extractLabeled))
.build();
}
/**
* As sequence data set dataset.
*
* @param sequenceType the sequence type
* @param featurizer the featurizer
* @return the dataset
*/
default Dataset<Sequence> asSequenceDataSet(
@NonNull AnnotationType sequenceType, @NonNull SequenceFeaturizer<Annotation> featurizer) {
return Dataset.sequence()
.type(getDataSetType())
.source(
asSequenceStream(sequenceType).map(seq -> featurizer.extractSequence(seq.iterator())))
.build();
}
/**
* As sequence data set dataset.
*
* @param labelFunction the label function
* @param featurizer the featurizer
* @return the dataset
*/
default Dataset<Sequence> asSequenceDataSet(
@NonNull Function<? super Annotation, String> labelFunction,
@NonNull SequenceFeaturizer<Annotation> featurizer) {
return Dataset.sequence()
.type(getDataSetType())
.source(
asSequenceStream(labelFunction).map(seq -> featurizer.extractSequence(seq.iterator())))
.build();
}
/**
* As regression data set dataset.
*
* @param featurizer the featurizer
* @return the dataset
*/
default Dataset<Instance> asRegressionDataSet(@NonNull Featurizer<HString> featurizer) {
return Dataset.regression()
.type(getDataSetType())
.source(stream().map(featurizer::extract))
.build();
}
