meridian.schema.model_consumer.ModelConsumer

Consumes a trained Meridian model and produces an Mmm proto.

meridian.schema.model_consumer.ModelConsumer(
    model_processors_classes: Sequence[ProcType]
)

model_processors A preset list of model processor types.
specs_to_processors_classes Returns a mapping of spec types to their corresponding processor types.

Methods

__call__

View source

__call__(
    mmm: model.Meridian,
    specs: Sequence[model_processor.Spec],
    model_id: str = ''
) -> mmm_pb.Mmm

Produces an Mmm schema for the model along with its analyses results.

Args
mmm A trained Meridian model. A trained model has its posterior distributions already sampled.
specs A sequence of specs that specify the analyses to run on the model. Specs of the same type will be grouped together and executed together by the corresponding model processor.
model_id An optional model identifier.

Returns
A proto containing the model kernel at rest and its analysis results.