pimlico.datatypes.sklearn module

Datatypes for storing and loading Scikit-learn models.

class SklearnModelWriter(base_dir, **kwargs)

Bases: pimlico.datatypes.base.PimlicoDatatypeWriter

write_model(model)

class SklearnModel(base_dir, pipeline, **kwargs)

Bases: pimlico.datatypes.base.PimlicoDatatype

Datatype for storing Scikit-learn models.

Very simple storage mechanism: we just pickle the model to a file. Instead of the standard Python pickle package, we use Joblib, which stores large data objects (especially Numpy arrays) more efficiently.

get_software_dependencies()

load_model()