def test_num_stack_levels():
assert ConfigBuilder().num_stack_levels(0).build() == dict(
num_stack_levels=0)
with pytest.raises(
AssertionError,
match=r"num_stack_levels must be greater or equal than zero"):
ConfigBuilder().num_stack_levels(-1).build()
def test_included_model_types():
expected_config = dict(excluded_model_types=[
'XT', 'KNN', 'GBM', 'CAT', 'XGB', 'NN_MXNET', 'NN_TORCH', 'LR',
'FASTAI', 'TRANSF', 'AG_TEXT_NN', 'AG_IMAGE_NN', 'FASTTEXT', 'VW'
])
actual_config = ConfigBuilder().included_model_types('RF').build()
assert actual_config == expected_config
actual_config = ConfigBuilder().included_model_types(['RF']).build()
assert actual_config == expected_config
actual_config = ConfigBuilder().included_model_types(['RF', 'RF']).build()
assert actual_config == expected_config
expected_config = dict(excluded_model_types=[
'XT', 'KNN', 'GBM', 'CAT', 'XGB', 'NN_MXNET', 'NN_TORCH', 'FASTAI',
'TRANSF', 'AG_TEXT_NN', 'AG_IMAGE_NN', 'FASTTEXT', 'VW'
])
actual_config = ConfigBuilder().included_model_types(['RF', 'LR']).build()
assert actual_config == expected_config
class CustomKNN(KNNModel):
pass
expected_config = dict(excluded_model_types=[
'XT', 'KNN', 'GBM', 'CAT', 'XGB', 'NN_MXNET', 'NN_TORCH', 'LR',
'FASTAI', 'TRANSF', 'AG_TEXT_NN', 'AG_IMAGE_NN', 'FASTTEXT', 'VW'
])
actual_config = ConfigBuilder().included_model_types([CustomKNN,
'RF']).build()
assert actual_config == expected_config
def test_excluded_model_types_invalid_option():
with pytest.raises(AssertionError,
match=r"unknown1 is not one of the valid models .*"):
ConfigBuilder().excluded_model_types('unknown1').build()
with pytest.raises(AssertionError,
match=r"unknown2 is not one of the valid models .*"):
ConfigBuilder().excluded_model_types(['unknown2']).build()
def test_time_limit():
expected_config = dict(time_limit=10)
actual_config = ConfigBuilder().time_limit(10).build()
assert actual_config == expected_config
expected_config = dict(time_limit=None)
actual_config = ConfigBuilder().time_limit(None).build()
assert actual_config == expected_config
def test_holdout_frac():
assert ConfigBuilder().holdout_frac(0).build() == dict(holdout_frac=0)
assert ConfigBuilder().holdout_frac(1).build() == dict(holdout_frac=1)
with pytest.raises(AssertionError,
match=r"holdout_frac must be between 0 and 1"):
ConfigBuilder().holdout_frac(-0.1).build()
with pytest.raises(AssertionError,
match=r"holdout_frac must be between 0 and 1"):
ConfigBuilder().holdout_frac(1.1).build()
def test_included_model_types_invalid_option():
with pytest.raises(
AssertionError,
match=
r"The following model types are not recognized: .'unknown1'. - use one of the valid models: .*"
):
ConfigBuilder().included_model_types('unknown1').build()
with pytest.raises(
AssertionError,
match=
r"The following model types are not recognized: .'unknown2', 'unknown3'. - use one of the valid models: .*"
):
ConfigBuilder().included_model_types(['RF', 'unknown2',
'unknown3']).build()
def test_presets_invalid_option():
with pytest.raises(
AssertionError,
match=
r"The following presets are not recognized: .'unknown1'. - use one of the valid presets: .*"
):
ConfigBuilder().presets('unknown1').build()
with pytest.raises(
AssertionError,
match=
r"The following presets are not recognized: .'unknown2', 'unknown3'. - use one of the valid presets: .*"
):
ConfigBuilder().presets(['best_quality', 'unknown2',
'unknown3']).build()
def test_hyperparameters_dict():
expected_config = dict(hyperparameters={'RF': {}})
actual_config = ConfigBuilder().hyperparameters({'RF': {}}).build()
assert actual_config == expected_config
class CustomKNN(KNNModel):
pass
expected_config = dict(hyperparameters={CustomKNN: [{}, {'prop': 42}]})
actual_config = ConfigBuilder().hyperparameters({
CustomKNN: [{}, {
'prop': 42
}]
}).build()
assert actual_config == expected_config
def test_excluded_model_types():
expected_config = dict(excluded_model_types=['RF'])
actual_config = ConfigBuilder().excluded_model_types('RF').build()
assert actual_config == expected_config
actual_config = ConfigBuilder().excluded_model_types(['RF']).build()
assert actual_config == expected_config
expected_config = dict(excluded_model_types=['LR', 'RF'])
actual_config = ConfigBuilder().excluded_model_types(['RF', 'LR']).build()
assert actual_config == expected_config
expected_config = dict(excluded_model_types=['RF'])
actual_config = ConfigBuilder().excluded_model_types(['RF', 'RF']).build()
assert actual_config == expected_config
def test_presets():
expected_config = dict(presets=['best_quality'])
actual_config = ConfigBuilder().presets('best_quality').build()
assert actual_config == expected_config
actual_config = ConfigBuilder().presets(['best_quality']).build()
assert actual_config == expected_config
expected_config = dict(presets=['best_quality', 'optimize_for_deployment'])
actual_config = ConfigBuilder().presets(
['best_quality', 'optimize_for_deployment']).build()
assert actual_config == expected_config
expected_config = dict(presets={'a': 42})
actual_config = ConfigBuilder().presets({'a': 42}).build()
assert actual_config == expected_config
def test_hyperparameter_tune_kwargs():
assert ConfigBuilder().hyperparameter_tune_kwargs('auto').build() == dict(
hyperparameter_tune_kwargs='auto')
assert ConfigBuilder().hyperparameter_tune_kwargs(
'random').build() == dict(hyperparameter_tune_kwargs='random')
assert ConfigBuilder().hyperparameter_tune_kwargs({
'props': 42
}).build() == dict(hyperparameter_tune_kwargs={'props': 42})
with pytest.raises(AssertionError,
match=r"unknown string must be one of .*"):
ConfigBuilder().hyperparameter_tune_kwargs('unknown').build()
with pytest.raises(
ValueError,
match=r"hyperparameter_tune_kwargs must be either str: .* or dict"
):
ConfigBuilder().hyperparameter_tune_kwargs(42).build()
def test_feature_generator_2():
config = (ConfigBuilder().feature_generator().enable_numeric_features(
False).enable_categorical_features(False).enable_datetime_features(
False).enable_text_special_features(False).
enable_text_ngram_features(False).enable_raw_text_features(
False).enable_vision_features(False).build().build())
assert config['feature_generator'].enable_numeric_features is False
assert config['feature_generator'].enable_categorical_features is False
assert config['feature_generator'].enable_datetime_features is False
assert config['feature_generator'].enable_text_special_features is False
assert config['feature_generator'].enable_text_ngram_features is False
assert config['feature_generator'].enable_raw_text_features is False
assert config['feature_generator'].enable_vision_features is False
text_gen = None
generators_classes = []
for gl in config['feature_generator'].generators:
for g in gl:
if isinstance(g, TextNgramFeatureGenerator):
text_gen = g
generators_classes.append(g.__class__.__name__)
assert text_gen is None
assert sorted(list(set(generators_classes))) == [
'AsTypeFeatureGenerator', 'DropUniqueFeatureGenerator',
'FillNaFeatureGenerator'
]
def test_hyperparameters__invalid_option():
with pytest.raises(
ValueError,
match=
r"hyperparameters must be either str: .* or dict with keys of .*"):
ConfigBuilder().hyperparameters(42).build()
with pytest.raises(AssertionError,
match=r"unknown is not one of the valid presets .*"):
ConfigBuilder().hyperparameters('unknown').build()
with pytest.raises(
AssertionError,
match=
r"The following model types are not recognized: .'unknown'. - use one of the valid models: .*"
):
ConfigBuilder().hyperparameters({'unknown': []}).build()
def test_feature_generator():
vectorizer = CountVectorizer(min_df=7,
ngram_range=(2, 3),
max_features=11,
dtype=np.uint8)
config = (ConfigBuilder().feature_generator().enable_numeric_features(
).enable_categorical_features().enable_datetime_features(
).enable_text_special_features().enable_text_ngram_features().
enable_raw_text_features().enable_vision_features().vectorizer(
vectorizer).text_ngram_params({
'vectorizer_strategy': 'both'
}).build().build())
assert config['feature_generator'].enable_numeric_features is True
assert config['feature_generator'].enable_categorical_features is True
assert config['feature_generator'].enable_datetime_features is True
assert config['feature_generator'].enable_text_special_features is True
assert config['feature_generator'].enable_text_ngram_features is True
assert config['feature_generator'].enable_raw_text_features is True
assert config['feature_generator'].enable_vision_features is True
text_gen = None
generators_classes = []
for gl in config['feature_generator'].generators:
for g in gl:
if isinstance(g, TextNgramFeatureGenerator):
text_gen = g
generators_classes.append(g.__class__.__name__)
print(generators_classes)
assert str(text_gen.vectorizer_default_raw) == str(vectorizer)
assert text_gen.vectorizer_strategy == 'both'
assert sorted(list(set(generators_classes))) == [
'AsTypeFeatureGenerator', 'CategoryFeatureGenerator',
'DatetimeFeatureGenerator', 'DropUniqueFeatureGenerator',
'FillNaFeatureGenerator', 'IdentityFeatureGenerator',
'IsNanFeatureGenerator', 'TextNgramFeatureGenerator',
'TextSpecialFeatureGenerator'
]
def test_ag_args_ensemble():
assert ConfigBuilder().ag_args_ensemble({
'param': 42
}).build() == dict(ag_args_ensemble={'param': 42})
def test_time_limit_invalid_option():
with pytest.raises(AssertionError,
match=r"time_limit must be greater than zero"):
ConfigBuilder().time_limit(-1).build()
def test_hyperparameters_str():
expected_config = dict(hyperparameters='very_light')
actual_config = ConfigBuilder().hyperparameters('very_light').build()
assert actual_config == expected_config
def test_refit_full():
assert ConfigBuilder().refit_full().build() == dict(refit_full=True)
assert ConfigBuilder().refit_full(False).build() == dict(refit_full=False)
assert ConfigBuilder().refit_full('best').build() == dict(
refit_full='best')
def test_use_bag_holdout():
assert ConfigBuilder().use_bag_holdout().build() == dict(
use_bag_holdout=True)
assert ConfigBuilder().use_bag_holdout(False).build() == dict(
use_bag_holdout=False)
def test_calibrate():
assert ConfigBuilder().calibrate().build() == dict(calibrate=True)
assert ConfigBuilder().calibrate(False).build() == dict(calibrate=False)
def test_save_space():
assert ConfigBuilder().save_space().build() == dict(save_space=True)
assert ConfigBuilder().save_space(False).build() == dict(save_space=False)
def test_keep_only_best():
assert ConfigBuilder().keep_only_best().build() == dict(
keep_only_best=True)
assert ConfigBuilder().keep_only_best(False).build() == dict(
keep_only_best=False)
def test_set_best_to_refit_full():
assert ConfigBuilder().set_best_to_refit_full().build() == dict(
set_best_to_refit_full=True)
assert ConfigBuilder().set_best_to_refit_full(False).build() == dict(
set_best_to_refit_full=False)
def test_auto_stack():
assert ConfigBuilder().auto_stack().build() == dict(auto_stack=True)
assert ConfigBuilder().auto_stack(False).build() == dict(auto_stack=False)
def test_ag_args_fit():
assert ConfigBuilder().ag_args_fit({
'param': 42
}).build() == dict(ag_args_fit={'param': 42})
def test_num_bag_sets():
assert ConfigBuilder().num_bag_sets(1).build() == dict(num_bag_sets=1)
with pytest.raises(AssertionError,
match=r"num_bag_sets must be greater than zero"):
ConfigBuilder().num_bag_sets(0).build()
