def test_linear():
hyperparameter_elm = {'reg': [10**i for i in range(-3, 4)]}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='LinearRegressor',
hyperparameter=hyperparameter_elm,
metric_list=['accuracy', 'rmse'])
def test_ols_regression():
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='OLS',
hyperparameter=None,
metric_list=['rmse'],
classification=False)
def test_linear_regression():
hyperparameter_elm = {'reg': [10**i for i in range(-2, 3)]}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='LinearRegressor',
hyperparameter=hyperparameter_elm,
metric_list=['rmse'],
classification=False)
def test_bagging():
hyperparameter_bagging = {'activation': ['sigmoid'],
'reg': [10 ** i for i in range(-1, 2)],
'hidden_neurons': [10],
'size': [5]}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='BaggingELM',
hyperparameter=hyperparameter_bagging,
metric_list=['accuracy', 'rmse'])
def test_kelm():
hyperparameter_kelm = {
'kernel': ['rbf', 'linear'],
'reg': [10**i for i in range(-1, 2)],
'gamma': [10]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='KernelELM',
hyperparameter=hyperparameter_kelm,
metric_list=['accuracy', 'rmse'])
def test_pl():
hyperparameter_elm = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'hidden_neurons': [10]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='ParallelLayerELM',
hyperparameter=hyperparameter_elm,
metric_list=['accuracy', 'rmse'])
def test_pcaldaelm():
hyperparameter_elm = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'pca_perc': [0.9]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='PCALDAELM',
hyperparameter=hyperparameter_elm,
metric_list=['accuracy'])
def test_bagging_regression():
hyperparameter_bagging = {'activation': ['sigmoid'],
'reg': [10 ** i for i in range(-1, 2)],
'hidden_neurons': [10],
'size': [5]}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='BaggingELM',
hyperparameter=hyperparameter_bagging,
metric_list=['rmse'],
classification=False)
def test_diverse_elm():
hyperparameter_div = {'activation': ['sigmoid'],
'reg': [10 ** i for i in range(-1, 2)],
'div': [10 ** i for i in range(-1, 2)],
'hidden_neurons': [10],
'size': [5]}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='DiverseELM',
hyperparameter=hyperparameter_div,
metric_list=['accuracy', 'rmse', 'diversity'])
def test_ncelm():
hyperparameter_inc = {'activation': ['sigmoid'],
'reg': [10 ** i for i in range(-2, 3)],
'lambda_': [10 ** i for i in range(-4, -2)],
'max_iter_': [5],
'hidden_neurons': [10],
'size': [5]}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='NegativeCorrelationELM',
hyperparameter=hyperparameter_inc,
metric_list=['accuracy', 'rmse'])
def test_adaboost():
hyperparameter_boost = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'hidden_neurons': [10],
'size': [5]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='AdaBoostELM',
hyperparameter=hyperparameter_boost,
metric_list=['accuracy'])
def test_pl_regression():
hyperparameter_elm = {
'activation': ['sigmoid'],
'reg': [0.001],
'hidden_neurons': [10]
}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='ParallelLayerELM',
hyperparameter=hyperparameter_elm,
metric_list=['rmse'],
classification=False)
def test_kelm_regression():
hyperparameter_kelm = {
'kernel': ['rbf', 'linear'],
'reg': [10**i for i in range(-1, 2)],
'gamma': [10]
}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='KernelELM',
hyperparameter=hyperparameter_kelm,
metric_list=['rmse'],
classification=False)
def test_pcaelm_regression():
hyperparameter_elm = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'pca_perc': [0.9]
}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='PCAELM',
hyperparameter=hyperparameter_elm,
metric_list=['rmse'],
classification=False)
def test_ncelm_regression():
hyperparameter_inc = {'activation': ['sigmoid'],
'reg': [10 ** i for i in range(-1, 2)],
'lambda_': [10 ** i for i in range(-2, -1)],
'max_iter_': [5],
'hidden_neurons': [10],
'size': [5]}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='NegativeCorrelationELM',
hyperparameter=hyperparameter_inc,
metric_list=['rmse'],
classification=False)
def test_ancelm():
hyperparameter_anc = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'hidden_neurons': [10],
'lambda_': [0.5, 1.0, 5.0],
'size': [5]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='AdaBoostNCELM',
hyperparameter=hyperparameter_anc,
metric_list=['accuracy'])
def test_adaboost_regression():
hyperparameter_boost = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'hidden_neurons': [10],
'size': [5]
}
with pytest.raises(ValueError):
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='AdaBoostELM',
hyperparameter=hyperparameter_boost,
metric_list=['rmse'],
classification=False)
def test_nn():
hyperparameter_nn = {
'max_iter': [200],
'activation': ['sigmoid'],
'hidden_neurons': [10 * i for i in range(1, 3)],
'solver': ['irprop'],
'batch_size': [50],
'learning_rate': [0.01]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='NeuralNetwork',
hyperparameter=hyperparameter_nn,
metric_list=['accuracy', 'rmse'])
def test_rnn():
hyperparameter_nn = {
'max_iter': [200],
'activation': ['sigmoid'],
'hidden_neurons': [5, 10, 15],
'solver': ['backpropagation'],
'batch_size': [75],
'learning_rate': [0.01]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='RandomNeuralNetwork',
hyperparameter=hyperparameter_nn,
metric_list=['accuracy', 'rmse'])
def test_nn_autoencoder():
hyperparameter_nn = {
'max_iter': [200],
'activation': ['sigmoid'],
'hidden_neurons': [2],
'solver': ['irprop'],
'batch_size': [50],
'learning_rate': [0.01]
}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='NeuralNetwork',
hyperparameter=hyperparameter_nn,
metric_list=['rmse'],
autoencoder=True)
def test_nc_nn():
hyperparameter_nn = {
'max_iter': [100],
'activation': ['sigmoid'],
'solver': ['irprop'],
'batch_size': [25],
'hidden_neurons': [5, 10],
'learning_rate': [0.01],
'size': [5],
'lambda_': [0.01]}
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='NegativeCorrelationNN',
hyperparameter=hyperparameter_nn,
metric_list=['accuracy', 'rmse'])
def test_rnn_regression():
hyperparameter_nn = {
'max_iter': [200],
'activation': ['sigmoid'],
'hidden_neurons': [5, 10, 15],
'solver': ['irprop'],
'batch_size': [150],
'learning_rate': [0.001]
}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='RandomNeuralNetwork',
hyperparameter=hyperparameter_nn,
metric_list=['rmse'],
classification=False)
def test_nc_nn_regression():
hyperparameter_nn = {'max_iter': [100],
'activation': ['sigmoid'],
'hidden_neurons': [5, 10],
'learning_rate': [0.001],
'solver': ['irprop'],
'batch_size': [50],
'size': [5],
'lambda_': [0.001]}
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='NegativeCorrelationNN',
hyperparameter=hyperparameter_nn,
metric_list=['rmse'],
classification=False)
def test_activation():
hyperparameter_elm = {
'activation': ['sigmoid'],
'reg': [10**i for i in range(-1, 2)],
'hidden_neurons': [10]
}
for activation in activation_dict.keys():
hyperparameter_elm['activation'] = [activation]
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='ELM',
hyperparameter=hyperparameter_elm,
metric_list=['rmse'],
classification=False)
if value_dict['rmse'] > 0.4:
raise ValueError(
'Activation function %s does not '
'provide good results', activation)
def test_nn_activation():
hyperparameter_nn = {
'max_iter': [100],
'activation': ['sigmoid'],
'hidden_neurons': [5],
'solver': ['irprop'],
'batch_size': [150],
'learning_rate': [0.001]
}
for activation in nn_activation_dict.keys():
hyperparameter_nn['activation'] = [activation]
value_dict = check_algorithm(folder='data_regression',
dataset='housing',
algorithm='NeuralNetwork',
hyperparameter=hyperparameter_nn,
metric_list=['rmse'],
classification=False)
if value_dict['rmse'] > 0.4:
raise ValueError(
'Activation function %s does not '
'provide good results', activation)
def test_ols():
value_dict = check_algorithm(folder='data',
dataset='iris',
algorithm='OLS',
hyperparameter=None,
metric_list=['accuracy', 'rmse'])
