def test_can_access_classifier_attribute(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models)
ensembler1.fit(self.X, self.Y)
est_type = getattr(ensembler1, "_estimator_type")
self.assertEqual(est_type, 'classifier')
def test_fit_predictproba(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models, voting='soft')
ensembler1.fit(self.X, self.Y)
pred = ensembler1.predict(self.X)
acs = accuracy_score(self.Y, pred)
print('Accuracy Score: ', acs)
def test_fit_base(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models)
ensembler1.fit_base(self.X, self.Y)
p1 = ensembler1.models[0].predict(self.X)
self.assertIsInstance(p1, numpy.ndarray)
p2 = ensembler1.models[1].predict(self.X)
self.assertIsInstance(p2, numpy.ndarray)
def test_load_saved_ensemble(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models)
ensembler1.fit(self.X, self.Y)
save_object('ensembler1.pkl', ensembler1)
# try loading
ens1 = load_object('ensembler1.pkl')
self.assertIsInstance(ens1, Ensembler)
def test_object_constructor(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models)
self.assertListEqual(ensembler1.models, models)
with self.assertRaises(AssertionError) as ae:
Ensembler(models, None, True)
with self.assertRaises(AssertionError) as ae:
Ensembler()
def test_saved_constructor(self):
rf = RandomForestClassifier(n_estimators=10)
rf.fit(self.X, self.Y)
save_object('rf.pkl', rf)
xg = XGBClassifier()
xg.fit(self.X, self.Y)
save_object('xg.pkl', xg)
es = Ensembler(modelpaths=[
'rf.pkl',
'xg.pkl'
])
es.fit(self.X, self.Y)
preds = es.predict(self.X)
self.assertIsInstance(preds, numpy.ndarray)
def test_intermediary_state_features(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models, voting='hard')
ensembler1.fit(self.X, self.Y)
meta_features1 = ensembler1.ensemble.named_steps['base_layer'].transform(self.X)
print(meta_features1.shape)
self.assertEqual(meta_features1.shape[0], self.X.shape[0])
self.assertEqual(meta_features1.shape[1], 2)
ensembler2 = Ensembler(models, voting='soft')
ensembler2.fit(self.X, self.Y)
meta_features2 = ensembler2.ensemble.named_steps['base_layer'].transform(self.X)
print(meta_features2.shape)
self.assertEqual(meta_features2.shape[0], self.X.shape[0])
self.assertEqual(meta_features2.shape[1], 6)
def test_compile(self):
models = [RandomForestClassifier(n_estimators=10), XGBClassifier()]
ensembler1 = Ensembler(models)
ensembler1.fit_base(self.X, self.Y)
ensemble = ensembler1.compile()
print(ensemble)