def test_scaler_minmax_1D(self):
ones = numpy.ones(10).reshape(1, -1)
for i in range(1, 5):
scaler = MinMaxScaler()
data = numpy.hstack([ones * i, ones * -i])
# Verify that normalization of data works as expected
norm = scaler.fit(data).transform(data)
self.assertEqual(norm[0].min(), 0)
self.assertEqual(norm[0].max(), 1)
def test_scaler_minmax(self):
ones = numpy.ones(10)
for i in range(1, 5):
scaler = MinMaxScaler()
data = numpy.concatenate([ones * i, ones * -i])
# Verify that normalization of data works as expected
norm = scaler.fit(data).transform(data)
self.assertEqual(norm.min(), 0)
self.assertEqual(norm.max(), 1)
def test_scaler_minmax_inverse(self):
ones = numpy.ones(10)
for i in range(1, 5):
scaler = MinMaxScaler()
data = numpy.concatenate([ones * i, ones * -i])
norm = scaler.fit(data).transform(data)
self.assertEqual(norm.min(), 0)
self.assertEqual(norm.max(), 1)
norm_inv = scaler.inverse_transform(norm)
self.assertListEqual(data.tolist(), norm_inv.tolist())
def test_scaler_minmax_2D_batches(self):
ones = numpy.ones(20).reshape(2, -1)
for i in range(1, 5):
scaler = MinMaxScaler(columns=[0, 1])
scaler = scaler.fit(ones * i)
scaler = scaler.fit(ones * -i)
# Verify that normalization of data works as expected
norm = scaler.transform(numpy.hstack([ones * i, ones * -i]))
self.assertEqual(norm[0].min(), 0)
self.assertEqual(norm[1].min(), 0)
self.assertEqual(norm[0].max(), 1)
self.assertEqual(norm[1].max(), 1)
def test_scaler_minmax_3D(self):
# Single column with 10 samples of an 8x8 matrix
ones = numpy.ones((10, 8, 8))
for i in range(1, 5):
scaler = MinMaxScaler()
# Wrap the matrix into an array to delineate single column
data = [numpy.vstack([ones * i, ones * -i])]
# Verify that normalization of data works as expected
norm = scaler.fit(data).transform(data)
self.assertEqual(norm[0].min(), 0)
self.assertEqual(norm[0].max(), 1)