logml.models.registry

Functions

list_eligible_models([objective, short])

List available models.

Classes

EligibleModels(*args, **kwargs)

Provides registry functionality for Model classes.

class logml.models.registry.EligibleModels(*args, **kwargs)

Bases: logml.common.BaseEligibleRegistrator

Provides registry functionality for Model classes.

LABEL_ATTRIBUTE = '__name__'

query_models(objective: Optional[logml.common.ModelingTask] = None, fi_attribute_exists: bool = False, exclude_tags: Optional[Iterable[str]] = None, include_tags: Optional[Iterable[str]] = None) → List[str]

Filter models be objective and tags. Default tags to exclude models = {‘meta’, ‘no_predict_proba’}

Parameters

objective – ModelingTask to filter models
fi_attribute_exists – if True, model has to have meaningful FE_MODEL_ATTRIBUTE
exclude_tags –

list of model.TAGS to exclude. If None, default tags are used:
(‘meta’, ‘no_predict_proba’, ‘exclude’). Pass empty list [] to not exclude any models.
include_tags – list of model.TAGS to include.

try_validate_model(model_id: Optional[str] = None, objective: Optional[logml.common.ModelingTask] = None): Same as validate model, but does not raise

validate_model(model_id: Optional[str] = None, objective: Optional[logml.common.ModelingTask] = None): Validates that model exists, and if objective provided - that the model matches the objective.

get_default_model(objective: logml.common.ModelingTask) → str: Returns default model for a given objective

mapping: Dict[str, Type]

logml.models.registry.list_eligible_models(objective: Optional[str] = None, short: bool = False) → None: List available models.

Modules

`logml.models.registry.dummy`
`logml.models.registry.linear`
`logml.models.registry.naive_bayes`
`logml.models.registry.neighbors`
`logml.models.registry.survival`	Survival models with sklearn api.
`logml.models.registry.svm`
`logml.models.registry.tree_based`