Skip to main content

pc.get_default_config

pc.get_default_config(dataset, targets=None, automl=False, hyperopt=False)

     Get the default config used to create a model. Can be populated with basic defaults,
     autoML defaults, or autoML with hyperopt defaults. This function can help as a good
     starting point for building a model config instead of having to create one from scratch.

Parameters:​

       dataset: str
       Dataset to generate default config from.

       targets: Union[str, List[str]], default None
       Target(s) in the dataset to use as output features.

       hyperopt: bool, default False
       Whether to generate the default config with autoML suggested parameters.

       hyperopt: bool, default False
       Whether to generate the default config with autoML suggested parameters and a hyperopt section.

Returns:​

       Config(dict)

Examples:​

Generate a default config from a dataset.

    config = pc.get_default_config(dataset, targets='Survived')
    config
#    > {'trainer': {'batch_size': 'auto',
#        'learning_rate': 'auto',
#        'learning_rate_scaling': 'sqrt',
#        'decay': True,
#        'decay_steps': 20000,
#        'decay_rate': 0.8,
#        'optimizer': {'type': 'adam'},
#        'tune_batch_size': {'type': 'bin_search', 'substitute_with_max': True}},
#        'input_features': [{'name': 'Pclass', 'column': 'Pclass', 'type': 'category'},
#        {'name': 'Sex', 'column': 'Sex', 'type': 'category'},
#        {'name': 'Age', 'column': 'Age', 'type': 'number'},
#        {'name': 'SibSp', 'column': 'SibSp', 'type': 'number'},
#        {'name': 'Parch', 'column': 'Parch', 'type': 'category'},
#        {'name': 'Fare', 'column': 'Fare', 'type': 'number'},
#        {'name': 'Cabin', 'column': 'Cabin', 'type': 'category'},
#        {'name': 'Embarked', 'column': 'Embarked', 'type': 'category'}],
#        'output_features': [{'name': 'Survived',
#        'column': 'Survived',
#        'type': 'binary'}],
#        'combiner': {'type': 'tabnet'}}