com.alpine.model.pack.preprocess
This should return a representative class from every jar that is needed to load the object during deserialization.
This should return a representative class from every jar that is needed to load the object during deserialization.
Default implementation returns the class of MLModel, and the class of the model implementation.
Used to identify this model when in a collection of models.
Used to identify this model when in a collection of models. Should be simple characters, so it can be used in a feature name.
identifier for the model.
Returns a streamlined version of this model, throwing away parts which are not required to produce the features with names in requiredOutputFeatureNames.
Returns a streamlined version of this model, throwing away parts which are not required to produce the features with names in requiredOutputFeatureNames.
Behaviour is undefined if requiredOutputFeatureNames is not actually a subset of outputFeatures.map(_columnName).
The resulting model may still have more output features than those required (i.e. for ClassificationRowModel implementations which always have the same output features, or for OneHotEncodingModel whose output features naturally group together).
names of the output features that the new model should produce.
A model which has the same functionality for those output features listed, but potentially fewer input features.
Model that creates output features that are polynomial combinations of the input features.
We use java.lang.Double for the type of the numeric values, because the scala Double type information is lost by scala/Gson and the deserialization fails badly for edge cases (e.g. Double.NaN).
If the exponents are a matrix
Then the transformation of a row (x1, x2, x3) will be