def test_pybrain_SoftMax_Tanh():
check_classifier(PyBrainClassifier(epochs=10, layers=[5, 2], hiddenclass=['TanhLayer', 'SoftmaxLayer'],
use_rprop=True),
**classifier_params)
check_regression(
PyBrainRegressor(epochs=2, layers=[10, 5, 2], hiddenclass=['TanhLayer', 'SoftmaxLayer', 'TanhLayer']),
**regressor_params)
def test_theanets_single_classification():
check_classifier(TheanetsClassifier(),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
check_classifier(TheanetsClassifier(layers=[]),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
check_classifier(TheanetsClassifier(layers=[20],
trainers=[{
'optimize': 'sgd',
'learning_rate': 0.3
}]),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
check_classifier(TheanetsClassifier(layers=[5, 5],
trainers=[{
'optimize': 'sgd',
'learning_rate': 0.3
}]),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
check_classifier(TheanetsClassifier(layers=[5, 5],
trainers=[{
'optimize': 'sgd',
'learning_rate': 0.3
}]),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
def test_theanets_multiclassification():
check_classifier(TheanetsClassifier(trainers=[{
'min_improvement': 0.1,
'learning_rate': 0.1
}]),
n_classes=4,
**classifier_params)
def test_complex_stacking_xgboost():
# Ada over kFold over xgboost
base_kfold = FoldingClassifier(base_estimator=XGBoostClassifier())
check_classifier(SklearnClassifier(
clf=AdaBoostClassifier(base_estimator=base_kfold, n_estimators=3)),
has_staged_pp=False,
has_importances=False)
def test_theanets_simple_stacking():
base_tnt = TheanetsClassifier()
check_classifier(SklearnClassifier(
clf=BaggingClassifier(base_estimator=base_tnt, n_estimators=3)),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
def test_simple_stacking_pybrain():
base_pybrain = PyBrainClassifier()
check_classifier(SklearnClassifier(
clf=BaggingClassifier(base_estimator=base_pybrain, n_estimators=3)),
has_staged_pp=False,
has_importances=False,
supports_weight=False)
def test_simple_stacking_nolearn():
# AdaBoostClassifier fails because sample_weight is not supported in nolearn
base_nl = NolearnClassifier()
check_classifier(SklearnClassifier(
clf=BaggingClassifier(base_estimator=base_nl, n_estimators=3)),
has_staged_pp=False,
has_importances=False,
supports_weight=False)
def test_theanets_multiple_classification():
check_classifier(
TheanetsClassifier(trainers=[{
'optimize': 'adadelta',
'min_improvement': 0.5
}, {
'optimize': 'nag'
}]), **classifier_params)
def test_theanets_configurations():
check_classifier(
TheanetsClassifier(layers=[20], scaler=False,
trainers=[{'optimize': 'nag', 'learning_rate': 0.3, 'min_improvement': 0.5}]),
**classifier_params)
check_classifier(
TheanetsClassifier(layers=[5, 5], trainers=[{'optimize': 'nag', 'learning_rate': 0.3, 'min_improvement': 0.5}]),
**classifier_params)
def test_complex_stacking_tmva():
# Ada over kFold over TMVA
base_kfold = FoldingClassifier(base_estimator=TMVAClassifier(),
random_state=13)
check_classifier(SklearnClassifier(
clf=AdaBoostClassifier(base_estimator=base_kfold, n_estimators=3)),
has_staged_pp=False,
has_importances=False)
def test_models():
for _ in range(3):
clf = CacheClassifier('clf', SGDClassifier(loss='log'))
check_classifier(clf, has_staged_pp=False, has_importances=False)
reg = CacheRegressor('reg', SGDRegressor())
check_regression(reg, has_staged_predictions=False, has_importances=False)
cache_helper.clear_cache()
def test_complex_stacking_mn():
# Ada over kFold over MatrixNet
base_kfold = FoldingClassifier(base_estimator=MatrixNetClassifier(
iterations=30))
check_classifier(SklearnClassifier(
clf=AdaBoostClassifier(base_estimator=base_kfold, n_estimators=3)),
has_staged_pp=False,
has_importances=False)
def test_pybrain_classification():
check_classifier(PyBrainClassifier(),
has_staged_pp=False,
has_importances=False,
supports_weight=False)
check_classifier(PyBrainClassifier(layers=[10, 10]),
has_staged_pp=False,
has_importances=False,
supports_weight=False)
def test_pybrain_Tanh():
check_classifier(PyBrainClassifier(layers=[10], hiddenclass=['TanhLayer']),
has_staged_pp=False,
has_importances=False,
supports_weight=False)
check_regression(PyBrainRegressor(layers=[10], hiddenclass=['TanhLayer']),
has_staged_predictions=False,
has_importances=False,
supports_weight=False)
def test_theanets_configurations():
check_classifier(
TheanetsClassifier(layers=[13], scaler=False,
trainers=[{'algo': 'nag', 'learning_rate': 0.1}]),
**classifier_params)
check_classifier(
TheanetsClassifier(layers=[5, 5], scaler='minmax',
trainers=[{'algo': 'adadelta', 'learning_rate': 0.1}]),
**classifier_params)
def test_theanets_multiple_classification():
check_classifier(TheanetsClassifier(trainers=[{
'optimize': 'adadelta'
}, {
'optimize': 'nag'
}]),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
def test_pretrain():
clf = TheanetsClassifier(trainers=[{
'optimize': 'pretrain',
'patience': 1,
'learning_rate': 0.1
}, {
'optimize': 'nag',
'patience': 1
}])
check_classifier(clf, **classifier_params)
def test_neurolab_stacking():
base_nlab = NeurolabClassifier(show=0,
layers=[],
epochs=N_EPOCHS2,
trainf=nl.train.train_rprop)
check_classifier(SklearnClassifier(
clf=BaggingClassifier(base_estimator=base_nlab, n_estimators=3)),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
def test_theanets_single_classification():
check_classifier(TheanetsClassifier(trainers=[{
'patience': 0
}]), **classifier_params)
check_classifier(
TheanetsClassifier(layers=[],
scaler='minmax',
trainers=[{
'patience': 0
}]), **classifier_params)
def test_pretrain():
clf = TheanetsClassifier(layers=[5, 5],
trainers=[{
'algo': 'pretrain',
'learning_rate': 0.1
}, {
'algo': 'nag',
'learning_rate': 0.1
}])
check_classifier(clf, **classifier_params)
def test_theanets_configurations():
check_classifier(
TheanetsClassifier(layers=[13], scaler=False,
trainers=[dict(algo='nag', learning_rate=0.1, **impatient)]),
**classifier_params)
check_classifier(
TheanetsClassifier(layers=[5, 5],
trainers=[dict(algo='adam', learning_rate=0.01, momentum=0.9)]
),
**classifier_params)
def test_models():
for _ in range(3):
clf = CacheClassifier('clf', SGDClassifier(loss='log'))
check_classifier(clf, has_staged_pp=False, has_importances=False)
reg = CacheRegressor('reg', SGDRegressor())
check_regression(reg,
has_staged_predictions=False,
has_importances=False)
cache_helper.clear_cache()
def test_theanets_single_classification():
check_classifier(TheanetsClassifier(),
supports_weight=False, has_staged_pp=False, has_importances=False)
check_classifier(TheanetsClassifier(layers=[]),
supports_weight=False, has_staged_pp=False, has_importances=False)
check_classifier(TheanetsClassifier(layers=[20], trainers=[{'optimize': 'sgd', 'learning_rate': 0.3}]),
supports_weight=False, has_staged_pp=False, has_importances=False)
check_classifier(TheanetsClassifier(layers=[5, 5], trainers=[{'optimize': 'sgd', 'learning_rate': 0.3}]),
supports_weight=False, has_staged_pp=False, has_importances=False)
check_classifier(TheanetsClassifier(layers=[5, 5], trainers=[{'optimize': 'sgd', 'learning_rate': 0.3}]),
supports_weight=False, has_staged_pp=False, has_importances=False)
def test_pybrain_Linear_MDLSTM():
check_classifier(
PyBrainClassifier(epochs=2,
layers=[10, 2],
hiddenclass=['LinearLayer', 'MDLSTMLayer']),
**classifier_params)
check_regression(
PyBrainRegressor(epochs=3,
layers=[10, 2],
hiddenclass=['LinearLayer', 'MDLSTMLayer']),
**regressor_params)
def test_theanets_configurations():
check_classifier(
TheanetsClassifier(
layers=[13],
scaler=False,
trainers=[dict(algo='nag', learning_rate=0.1, **impatient)]),
**classifier_params)
check_classifier(
TheanetsClassifier(
layers=[5, 5],
trainers=[dict(algo='adam', learning_rate=0.01, momentum=0.9)]),
**classifier_params)
def test_pybrain_SoftMax_Tanh():
check_classifier(
PyBrainClassifier(epochs=10,
layers=[5, 2],
hiddenclass=['TanhLayer', 'SoftmaxLayer'],
use_rprop=True), **classifier_params)
check_regression(
PyBrainRegressor(
epochs=2,
layers=[10, 5, 2],
hiddenclass=['TanhLayer', 'SoftmaxLayer', 'TanhLayer']),
**regressor_params)
def test_nolearn_classification():
cl = NolearnClassifier()
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False,
supports_weight=False)
cl = NolearnClassifier(layers=[])
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False,
supports_weight=False)
def test_theanets_configurations():
check_classifier(
TheanetsClassifier(layers=[20],
scaler=False,
trainers=[{
'optimize': 'nag',
'learning_rate': 0.3,
'min_improvement': 0.5
}]), **classifier_params)
check_classifier(
TheanetsClassifier(layers=[5, 5],
trainers=[{
'optimize': 'nag',
'learning_rate': 0.3,
'min_improvement': 0.5
}]), **classifier_params)
def test_neurolab_single_classification():
check_classifier(NeurolabClassifier(show=0,
layers=[],
epochs=N_EPOCHS2,
trainf=nl.train.train_rprop),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
check_classifier(NeurolabClassifier(net_type='single-layer',
cn='auto',
show=0,
epochs=N_EPOCHS2,
trainf=nl.train.train_delta),
supports_weight=False,
has_staged_pp=False,
has_importances=False)
def test_neurolab_single_classification():
check_classifier(NeurolabClassifier(layers=[], epochs=N_EPOCHS2, trainf=None),
**classifier_params)
check_classifier(NeurolabClassifier(layers=[2], epochs=N_EPOCHS2),
**classifier_params)
check_classifier(NeurolabClassifier(layers=[1, 1], epochs=N_EPOCHS2),
**classifier_params)
def test_tmva():
# check classifier
check_classifier(TMVAClassifier(),
check_instance=True,
has_staged_pp=False,
has_importances=False)
cl = TMVAClassifier(method='kSVM',
Gamma=0.25,
Tol=0.001,
sigmoid_function='identity')
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False)
cl = TMVAClassifier(method='kCuts',
FitMethod='GA',
EffMethod='EffSel',
sigmoid_function='sig_eff=0.9')
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False)
# check regressor, need to run twice to check for memory leak.
for i in range(2):
check_regression(TMVARegressor(),
check_instance=True,
has_staged_predictions=False,
has_importances=False)
def test_neurolab_single_classification():
check_classifier(
NeurolabClassifier(layers=[], epochs=N_EPOCHS2, trainf=None),
**classifier_params)
check_classifier(NeurolabClassifier(layers=[2], epochs=N_EPOCHS2),
**classifier_params)
check_classifier(NeurolabClassifier(layers=[1, 1], epochs=N_EPOCHS2),
**classifier_params)
def test_sklearn_classification():
check_classifier(
SklearnClassifier(clf=AdaBoostClassifier(n_estimators=10)))
check_classifier(
SklearnClassifier(clf=AdaBoostClassifier(n_estimators=10)),
n_classes=3)
check_classifier(
SklearnClassifier(clf=GradientBoostingClassifier(n_estimators=10)))
def test_pybrain_classification():
clf = PyBrainClassifier(epochs=2)
check_classifier(clf, **classifier_params)
check_classifier(
PyBrainClassifier(epochs=-1, continue_epochs=1, layers=[]),
**classifier_params)
check_classifier(PyBrainClassifier(epochs=2, layers=[5, 2]),
**classifier_params)
def test_nolearn_classification():
cl = NolearnClassifier()
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False, supports_weight=False)
cl = NolearnClassifier(layers=[])
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False, supports_weight=False)
cl = NolearnClassifier(layers=[5, 5])
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False, supports_weight=False)
def test_sklearn_classification():
# supports weights
check_classifier(
SklearnClassifier(clf=AdaBoostClassifier(n_estimators=10)))
check_classifier(
SklearnClassifier(clf=AdaBoostClassifier(n_estimators=10)),
n_classes=3)
# doesn't support weights
check_classifier(
SklearnClassifier(clf=GradientBoostingClassifier(n_estimators=10)),
supports_weight=False)
def test_tmva():
# check classifier
check_classifier(TMVAClassifier(), check_instance=True, has_staged_pp=False, has_importances=False)
cl = TMVAClassifier(method='kSVM', Gamma=0.25, Tol=0.001, sigmoid_function='identity')
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False)
cl = TMVAClassifier(method='kCuts', FitMethod='GA', EffMethod='EffSel', sigmoid_function='sig_eff=0.9')
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False)
# check regressor, need to run twice to check for memory leak.
for i in range(2):
check_regression(TMVARegressor(), check_instance=True, has_staged_predictions=False, has_importances=False)
def test_tmva():
# check classifier
factory_options = "Silent=True:V=False:DrawProgressBar=False"
cl = TMVAClassifier(factory_options=factory_options, method='kBDT', NTrees=10)
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False)
cl = TMVAClassifier(factory_options=factory_options, method='kSVM', Gamma=0.25, Tol=0.001,
sigmoid_function='identity')
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False)
cl = TMVAClassifier(factory_options=factory_options, method='kCuts',
FitMethod='GA', EffMethod='EffSel', sigmoid_function='sig_eff=0.9')
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False)
# check regressor, need to run twice to check for memory leak.
for i in range(2):
check_regression(TMVARegressor(factory_options=factory_options, method='kBDT', NTrees=10), check_instance=True,
has_staged_predictions=False, has_importances=False)
def test_tmva():
# check classifier
factory_options = "Silent=True:V=False:DrawProgressBar=False"
cl = TMVAClassifier(factory_options=factory_options,
method='kBDT',
NTrees=10)
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False)
cl = TMVAClassifier(factory_options=factory_options,
method='kSVM',
Gamma=0.25,
Tol=0.001,
sigmoid_function='identity')
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False)
cl = TMVAClassifier(factory_options=factory_options,
method='kCuts',
FitMethod='GA',
EffMethod='EffSel',
sigmoid_function='sig_eff=0.9')
check_classifier(cl,
check_instance=True,
has_staged_pp=False,
has_importances=False)
# check regressor, need to run twice to check for memory leak.
for i in range(2):
check_regression(TMVARegressor(factory_options=factory_options,
method='kBDT',
NTrees=10),
check_instance=True,
has_staged_predictions=False,
has_importances=False)
def test_simple_stacking_xgboost():
base_xgboost = XGBoostClassifier()
classifier = SklearnClassifier(clf=AdaBoostClassifier(base_estimator=base_xgboost, n_estimators=3))
check_classifier(classifier,
has_staged_pp=False)
def test_xgboost():
check_classifier(XGBoostClassifier(n_estimators=20), n_classes=2)
check_classifier(XGBoostClassifier(n_estimators=20), n_classes=4)
check_regression(XGBoostRegressor(n_estimators=20))
def test_theanets_single_classification():
check_classifier(TheanetsClassifier(trainers=[{'patience': 0}]), **classifier_params)
check_classifier(TheanetsClassifier(layers=[], scaler='minmax',
trainers=[{'patience': 0}]), **classifier_params)
def test_neurolab_multiclassification():
check_classifier(NeurolabClassifier(layers=[10], epochs=N_EPOCHS4, trainf=nl.train.train_rprop),
n_classes=4, **classifier_params)
def test_neurolab_stacking():
base_nlab = NeurolabClassifier(layers=[], epochs=N_EPOCHS2 * 2, trainf=nl.train.train_rprop)
base_bagging = BaggingClassifier(base_estimator=base_nlab, n_estimators=3)
check_classifier(SklearnClassifier(clf=base_bagging), **classifier_params)
def test_pretrain():
clf = TheanetsClassifier(trainers=[{'optimize': 'pretrain', 'patience': 1, 'learning_rate': 0.1},
{'optimize': 'nag', 'patience': 1}])
check_classifier(clf, **classifier_params)
def test_theanets_multiclassification():
check_classifier(TheanetsClassifier(trainers=[{'patience': 0}]), n_classes=4, **classifier_params)
def test_theanets_simple_stacking():
base_tnt = TheanetsClassifier(trainers=[{'min_improvement': 0.1}])
base_bagging = BaggingClassifier(base_estimator=base_tnt, n_estimators=3)
check_classifier(SklearnClassifier(clf=base_bagging), **classifier_params)
def test_nolearn_multiple_classification():
cl = NolearnClassifier()
check_classifier(cl, check_instance=True, has_staged_pp=False, has_importances=False, supports_weight=False,
n_classes=4)
def test_pybrain_multi_classification():
check_classifier(PyBrainClassifier(), n_classes=4, **classifier_params)
def test_pybrain_classification():
clf = PyBrainClassifier(epochs=2)
check_classifier(clf, **classifier_params)
check_classifier(PyBrainClassifier(epochs=-1, continue_epochs=1, layers=[]), **classifier_params)
check_classifier(PyBrainClassifier(epochs=2, layers=[5, 2]), **classifier_params)
def test_sklearn_classification():
# supports weights
check_classifier(SklearnClassifier(clf=AdaBoostClassifier(n_estimators=10)))
check_classifier(SklearnClassifier(clf=AdaBoostClassifier(n_estimators=10)), n_classes=3)
# doesn't support weights
check_classifier(SklearnClassifier(clf=GradientBoostingClassifier(n_estimators=10)), supports_weight=False)
def test_simple_stacking_pybrain():
base_pybrain = PyBrainClassifier(epochs=2)
base_bagging = BaggingClassifier(base_estimator=base_pybrain, n_estimators=3)
check_classifier(SklearnClassifier(clf=base_bagging), **classifier_params)
def test_simple_stacking_sklearn():
base_sk = AdaBoostClassifier(n_estimators=30)
check_classifier(SklearnClassifier(clf=AdaBoostClassifier(base_estimator=base_sk, n_estimators=3)))
def test_pybrain_Linear_MDLSTM():
check_classifier(PyBrainClassifier(epochs=2, layers=[10, 2], hiddenclass=['LinearLayer', 'MDLSTMLayer']),
**classifier_params)
check_regression(PyBrainRegressor(epochs=3, layers=[10, 2], hiddenclass=['LinearLayer', 'MDLSTMLayer']),
**regressor_params)
def test_simple_stacking_tmva():
base_tmva = TMVAClassifier()
check_classifier(SklearnClassifier(clf=BaggingClassifier(base_estimator=base_tmva, n_estimators=3, random_state=13)),
has_staged_pp=False, has_importances=False)
def test_complex_stacking_tmva():
# Ada over kFold over TMVA
base_kfold = FoldingClassifier(base_estimator=TMVAClassifier(), random_state=13)
check_classifier(SklearnClassifier(clf=AdaBoostClassifier(base_estimator=base_kfold, n_estimators=3)),
has_staged_pp=False, has_importances=False)
def test_complex_stacking_xgboost():
# Ada over kFold over xgboost
base_kfold = FoldingClassifier(base_estimator=XGBoostClassifier())
check_classifier(SklearnClassifier(clf=AdaBoostClassifier(base_estimator=base_kfold, n_estimators=3)),
has_staged_pp=False, has_importances=False)
def test_simple_stacking_nolearn():
# AdaBoostClassifier fails because sample_weight is not supported in nolearn
base_nl = NolearnClassifier()
check_classifier(SklearnClassifier(clf=BaggingClassifier(base_estimator=base_nl, n_estimators=3)),
has_staged_pp=False, has_importances=False, supports_weight=False)
def test_theanets_multiple_classification():
check_classifier(TheanetsClassifier(trainers=[{'optimize': 'adadelta', 'min_improvement': 0.5}, {'optimize': 'nag'}]),
**classifier_params)
