def test_store(__model_object__):
""""""
model1 = __model_object__
model1.store()
assert os.path.isfile(''.join([model1.run_name, '_data.csv']))
assert os.path.isfile(''.join([model1.run_name, '_attributes.json']))
delete_files(model1.run_name)
def test_nn_hypertune(__data_split_model__):
model1 = __data_split_model__
model1.reg()
model1.make_grid()
model1.hyperTune()
assert type(model1.params) is dict
assert os.path.isfile(''.join([model1.run_name,
'.h5'])) # Check for PVA graphs
delete_files(model1.run_name)
def test_train_reg(__data_split_model__):
model1 = __data_split_model__
model1.reg()
model1.make_grid()
model1.hyperTune()
model1.train_reg()
assert type(model1.predictions) is pandas.core.frame.DataFrame
assert type(model1.predictions_stats) is dict
__assert_results__(model1.predictions_stats)
delete_files(model1.run_name)
def test_pickle(__run_all__without_analyze):
""""""
model1 = __run_all__without_analyze
model1.pickle_model()
from sklearn.metrics import mean_squared_error, r2_score
model2 = unpickle_model(''.join([model1.run_name, '.pkl']))
model2.run()
# Make predictions
predictions = model2.estimator.predict(model2.test_features)
# Dataframe for replicate_model
pva = pd.DataFrame([], columns=['actual', 'predicted'])
pva['actual'] = model2.test_target
pva['predicted'] = predictions
assert r2_score(pva['actual'], pva['predicted']) < 0.9
assert mean_squared_error(pva['actual'], pva['predicted']) > 0.2
assert np.sqrt(mean_squared_error(pva['actual'], pva['predicted'])) > 0.2
delete_files(model1.run_name)
def test_pva_graph(__run_all__):
model1 = __run_all__
assert os.path.isfile(''.join([model1.run_name,
'_PVA.png'])) # Check for PVA graphs
delete_files(model1.run_name)
def test_var_importance(__run_all__):
model1 = __run_all__
assert os.path.isfile(''.join([model1.run_name, '_importance-graph.png']))
delete_files(model1.run_name)