def test_set_params(self):
# Initialize
estimator_class = LinearRegression
amount_of_steps = 30
estimator_kwargs = {'fit_intercept': False}
sort_col = 'date'
steps_estimator = StepsSelectorEstimator(estimator_class,
amount_of_steps,
estimator_kwargs, sort_col)
# Set params
estimator_class = LogisticRegression
amount_of_steps = 40
estimator_kwargs = {'fit_intercept': True}
sort_col = 'month'
steps_estimator.set_params(
estimator_class=LogisticRegression,
amount_of_steps=amount_of_steps,
estimator_kwargs=estimator_kwargs,
sort_col=sort_col,
)
# Assert deep
params = steps_estimator.get_params()
self.assertEqual(len(params), 4)
self.assertEqual(params['estimator_class'], Classer(estimator_class))
self.assertEqual(params['amount_of_steps'], amount_of_steps)
self.assertEqual(params['estimator_kwargs'], estimator_kwargs)
self.assertEqual(params['sort_col'], sort_col)
def test_estimate(self):
# Fake data with a v shape. First 30 drop. Last 30 Increase.
x = list(range(30)) + list(range(30))
y = pd.Series(list(range(29, -1, -1)) + list(range(30)))
X = pd.DataFrame({'x': x})
# Initialize
estimator_class = LinearRegression
steps_estimator = StepsSelectorEstimator(estimator_class, 30)
# Fit
steps_estimator.fit(X, y)
# Predict
y_pred = steps_estimator.predict(X)
# Assert that only the increasing slope was fitted
for i in range(60):
self.assertAlmostEqual(y_pred[i], i % 30)
def test_get_params_shallow(self):
# Initialize
estimator_class = LinearRegression
amount_of_steps = 30
estimator_kwargs = {'fit_intercept': False}
sort_col = 'date'
steps_estimator = StepsSelectorEstimator(estimator_class,
amount_of_steps,
estimator_kwargs, sort_col)
# Assert shallow
params = steps_estimator.get_params(deep=False)
self.assertEqual(params['estimator_class'], Classer(estimator_class))
self.assertEqual(params['amount_of_steps'], amount_of_steps)
self.assertEqual(params['estimator_kwargs'], estimator_kwargs)
self.assertTrue(params['estimator_kwargs'] is estimator_kwargs)
self.assertEqual(params['sort_col'], sort_col)
def test_fit_unsorted(self):
# Fake data with a v shape. First 30 drop. Last 30 Increase.
x = list(range(30)) + list(range(30))
y = pd.Series(list(range(29, -1, -1)) + list(range(30)))
X = pd.DataFrame({'x': x})
# Initialize
estimator_class = LinearRegression
steps_estimator = StepsSelectorEstimator(estimator_class, 30)
steps_estimator._estimator = MagicMock()
# Fit
steps_estimator.fit(X, y)
# Assert
args, kwargs = steps_estimator._estimator.fit.call_args
self.assertEqual(len(args), 2)
self.assertEqual(len(kwargs), 0)
X_call, y_call = args
self.assertEqual(X.iloc[-30:].values.tolist(), X_call.values.tolist())
self.assertEqual(y.iloc[-30:].values.tolist(), y_call.values.tolist())
def test_repr(self):
# Initialize
estimator_class = LinearRegression
amount_of_steps = 30
estimator_kwargs = {'fit_intercept': False}
sort_col = 'date'
steps_estimator = StepsSelectorEstimator(estimator_class,
amount_of_steps,
estimator_kwargs, sort_col)
# Assert
expected = ('StepsSelectorEstimator('
'estimator_class=Classer(LinearRegression), '
'amount_of_steps=30, '
'estimator_kwargs={\'fit_intercept\': False})')
self.assertEqual(str(steps_estimator), expected)
def test_init(self):
# Arguments
estimator_class = LinearRegression
amount_of_steps = 30
estimator_kwargs = {'fit_intercept': False}
sort_col = 'date'
# Initialize
se = StepsSelectorEstimator(estimator_class, amount_of_steps,
estimator_kwargs, sort_col)
# Asserts
self.assertEqual(se._estimator.__class__, estimator_class)
self.assertFalse(se._estimator.fit_intercept)
self.assertEqual(se.sort_col, sort_col)
self.assertEqual(se.estimator_kwargs, estimator_kwargs)
self.assertTrue(se.estimator_kwargs is estimator_kwargs)