com.spotify.featran

Type class for collections to extract features from.

Type class for types to build feature into.

Encapsulate features extracted from a FeatureSpec.

Companion object for FeatureSpec.

Encapsulate specification for feature extraction and transformation.

Companion to FlatReader. Sometimes for serialization and compatability reasons it is better to write out data in an intermediate format such as JSON or tf.examples to interface with storage or other systems. This class uses the functions internal to a spec to write out the data into a new flat format.

Sometimes it is useful to store the features in an intermediate state in normally a flat version like Examples or maybe JSON. This makes it easier to interface with other systems.

TypeClass that is used to read data from flat files. The requirement is that each feature comes from the same type and can be looked up by name.

TypeClass for implementing the writer to a flat format keyed by name

Type class for floating point primitives.

Encapsulate features extracted from a MultiFeatureSpec. Allows separation back into specs by names or vectors.

Companion object for MultiFeatureSpec.

Wrapper for FeatureSpec that allows for combination and separation of different specs.

A SparseArray with names of non-zero entries.

Encapsulate RecordExtractor for extracting individual records.

A sparse representation of an array using two arrays for indices and values of non-zero entries.