def test_bigger_len(self):
X, y = self.get_dataset(n_samples=9888, n_positives=10)
gen = BalancedGenerator(X,
y,
positive_sample_perc=0.5,
np_ratio=1,
negative_perc=1)
self.assertEqual(gen.__len__(), np.ceil(9878 / 5))
def test_len(self):
X, y = self.get_dataset(n_samples=30, n_positives=5)
gen = BalancedGenerator(X,
y,
positive_sample_perc=0.1,
np_ratio=1,
negative_perc=1)
self.assertEqual(gen.__len__(), 25)
def test_negative_perc_len(self):
n_samples = 1000000
n_positives = 300
X, y = self.get_dataset(n_samples, n_positives)
gen = BalancedGenerator(X,
y,
positive_sample_perc=0.6,
np_ratio=1,
negative_perc=0.7)
self.assertEqual(gen.__len__(),
np.ceil(((n_samples - n_positives) * 0.7) / 180))
def test_bootstrapping_len(self):
n_samples = 1000000
n_positives = 300
X, y = self.get_dataset(n_samples, n_positives)
gen = BalancedGenerator(X,
y,
positive_sample_perc=1,
np_ratio=1,
negative_perc=1)
self.assertEqual(gen.__len__(), np.ceil(
(n_samples - n_positives) / 300))
def test_get_item(self):
n_samples = 100
n_positives = 10
X, y = self.get_dataset(n_samples, n_positives)
dataset = list(zip(X, y))
gen = BalancedGenerator(X,
y,
positive_sample_perc=1,
np_ratio=1,
negative_perc=1)
batch_X, batch_y = gen.__getitem__(0)
batch = list(zip(batch_X, batch_y))
self.assertDatasetContainsBatch(dataset, batch)
def test_get_all_items(self):
print("started test")
n_samples = 100
n_positives = 10
X, y = self.get_dataset(n_samples, n_positives)
dataset = list(zip(X, y))
gen = BalancedGenerator(X,
y,
positive_sample_perc=1,
np_ratio=1,
negative_perc=0.5)
for i in range(gen.__len__()):
batch_X, batch_y = gen.__getitem__(i)
batch = list(zip(batch_X, batch_y))
self.assertDatasetContainsBatch(dataset, batch)