def test_concatenate_good(self):
a = np.random.rand(5, 2)
b = np.random.rand(5, 2)
x = np.random.rand(5, 2)
y = np.random.rand(5, 2)
c1 = [x, y]
n1 = en.NumpyList(c1)
c2 = [a, b]
n2 = en.NumpyList(c2)
l1, l2 = n1.concat(n2).lists
self.assertEqual((l1 == np.concatenate([x, a])).all(), True,
f'Concatenate of first list failed')
self.assertEqual((l2 == np.concatenate([y, b])).all(), True,
f'Concatenate of first list failed')
def test_shuffle(self):
x = np.arange(5)
y = np.arange(5)
c = [x, y]
n = en.NumpyList(c)
xs, ys = n.shuffle().lists
self.assertEqual(
np.array(xs == ys).all(), True,
f'Rows not equal after 1st shuffle')
n = en.NumpyList(c)
xs, ys = n.shuffle().lists
self.assertEqual(
np.array(xs == ys).all(), True,
f'Rows not equal after 2nd shuffle')
def test_lists(self):
x = np.random.rand(5, 2)
y = np.random.rand(5, 2)
c = [x, y]
n = en.NumpyList(c)
self.assertEqual(len(n.lists), len(c),
f'Number of lists does not add up {len(n.lists)}')
self.assertEqual((n.lists[0] == x).all(), True, f'Lists not equal')
self.assertEqual((n.lists[1] == y).all(), True, f'Lists not equal')
def test_remove(self):
x = np.random.rand(5, 2)
y = np.random.rand(5, 2)
c = [x, y]
n = en.NumpyList(c)
n.remove(1)
self.assertEqual(n.number_of_lists,
len([x, y]) - 1,
f'Length not reduced after pop {n.number_of_lists}')
with self.assertRaises(en.NumpyListException):
n.remove(1)
def test_sample(self):
x = np.arange(5)
y = np.arange(5)
c = [x, y]
n = en.NumpyList(c)
xs, ys = n.sample(3).lists
self.assertEqual(
np.array(xs == ys).all(), True, f'Rows not equal after 1st sample')
xs, ys = n.sample(2).lists
self.assertEqual(
np.array(xs == ys).all(), True, f'Rows not equal after 2nd sample')
with self.assertRaises(en.NumpyListException):
_ = n.sample(len(x) + 1)
def test_pop(self):
x = np.random.rand(5, 2)
y = np.random.rand(5, 2)
c = [x, y]
n = en.NumpyList(c)
a = n.pop(1)
self.assertEqual(n.number_of_lists,
len([x, y]) - 1,
f'Length not reduced after pop {n.number_of_lists}')
self.assertEqual((y == a).all(), True,
f'Popped list not equal to original')
with self.assertRaises(en.NumpyListException):
n.pop(1)
def test_slice_bad(self):
x = np.random.rand(5, 2)
y = np.random.rand(5, 2)
c = [x, y]
n = en.NumpyList(c)
with self.assertRaises(en.NumpyListException):
_ = n[5]
with self.assertRaises(en.NumpyListException):
_ = n[-6]
with self.assertRaises(en.NumpyListException):
_ = n[0:6]
with self.assertRaises(en.NumpyListException):
_ = n[-1:5]
def test_concatenate_bad(self):
# Different length
a = np.random.rand(5, 2)
b = np.random.rand(5, 2)
x = np.random.rand(5, 3)
y = np.random.rand(5, 3)
c1 = [x, y]
n1 = en.NumpyList(c1)
c2 = [a, b]
n2 = en.NumpyList(c2)
with self.assertRaises(en.NumpyListException):
_ = n1.concat(n2).lists
# Different dimensions
a = np.random.rand(5, 2)
b = np.random.rand(5, 2)
x = np.random.rand(5, 2, 1)
y = np.random.rand(5, 2, 1)
c1 = [x, y]
n1 = en.NumpyList(c1)
c2 = [a, b]
n2 = en.NumpyList(c2)
with self.assertRaises(en.NumpyListException):
_ = n1.concat(n2).lists
def test_unique(self):
x = np.random.rand(5, 2)
y = np.random.randint(32, size=(5, 2))
c = [x, y]
n = en.NumpyList(c)
vl_1, cn_1 = np.unique(y, return_counts=True)
vl_1, cn_1 = list(vl_1), list(cn_1)
vl_2, cn_2 = n.unique(1)
self.assertListEqual(
vl_1, vl_2,
f'Unique values not correct. Got {vl_2}. Expected {vl_1}')
self.assertListEqual(
cn_1, cn_2,
f'Unique counts not correct. Got {cn_2}. Expected {cn_1}')
with self.assertRaises(en.NumpyListException):
_, _ = n.unique(0)
def test_slice_good(self):
x = np.random.rand(5, 2)
y = np.random.rand(5, 2)
c = [x, y]
n = en.NumpyList(c)
x0, y0 = n[0].lists
self.assertEqual(
np.array(x0 == x[0]).all(), True,
f'First entries do not match {x0}, {x[0]}')
self.assertEqual(
np.array(y0 == y[0]).all(), True,
f'First entries do not match {y0}, {y[0]}')
x1, y1 = n[1].lists
self.assertEqual(
np.array(x1 == x[1]).all(), True,
f'Second entries do not match {x1}, {x[1]}')
self.assertEqual(
np.array(y1 == y[1]).all(), True,
f'Second entries do not match {y1}, {y[1]}')
xf, yf = n[0:5].lists
self.assertEqual(
np.array(xf == x).all(), True,
f'All entries do not match {xf}, {x}')
self.assertEqual(
np.array(yf == y).all(), True,
f'All entries do not match {yf}, {y}')
xm, ym = n[1:4].lists
self.assertEqual(
np.array(xm == x[1:4]).all(), True,
f'Mid entries do not match {xf}, {x[1:4]}')
self.assertEqual(
np.array(ym == y[1:4]).all(), True,
f'Mid entries do not match {yf}, {y[1:4]}')
xl, yl = n[4].lists
self.assertEqual(
np.array(xl == x[-1]).all(), True,
f'Last entries do not match {xl}, {x[-1]}')
self.assertEqual(
np.array(yl == y[-1]).all(), True,
f'Last entries do not match {yl}, {y[-1]}')
xl, yl = n[-1].lists
self.assertEqual(
np.array(xl == x[-1]).all(), True,
f'Last entries do not match {xl}, {x[-1]}')
self.assertEqual(
np.array(yl == y[-1]).all(), True,
f'Last entries do not match {yl}, {y[-1]}')
def test_split(self):
length = 20
width = 5
a = np.random.rand(length, width)
b = np.random.rand(length, width)
c1 = [a, b]
n1 = en.NumpyList(c1)
# Split more data than available
with self.assertRaises(en.NumpyListException):
n1.split_time(1, length)
test_s, val_s = 2, 1
train, val, test = n1.split_time(val_s, test_s)
self.assertEqual(len(train.lists), len(c1),
f'Number of train lists changed {len(train.lists)}')
self.assertEqual(len(val.lists), len(c1),
f'Number of val lists changed {len(val.lists)}')
self.assertEqual(len(test.lists), len(c1),
f'Number of test lists changed {len(test.lists)}')
self.assertEqual((train.lists[0].shape[1]), width,
f'Width changed {train.lists[0].shape[1]}')
self.assertEqual((train.lists[1].shape[1]), width,
f'Width changed {train.lists[1].shape[1]}')
self.assertEqual((val.lists[0].shape[1]), width,
f'Width changed {val.lists[0].shape[1]}')
self.assertEqual((val.lists[1].shape[1]), width,
f'Width changed {val.lists[1].shape[1]}')
self.assertEqual((test.lists[0].shape[1]), width,
f'Width changed {test.lists[0].shape[1]}')
self.assertEqual((test.lists[1].shape[1]), width,
f'Width changed {test.lists[1].shape[1]}')
self.assertEqual(
len(val), val_s,
f'Expected Validation to be size {val_s}. Got {len(val)}')
self.assertEqual(
len(test), test_s,
f'Expected Test to be of size {test_s}. Got {len(test)}')
self.assertEqual(len(train), length - val_s - test_s,
f'Unexpected Length training. {length-val_s-test_s}')
def test_creation_base(self):
x = np.arange(10)
y = np.arange(10)
c = [x, y]
n = en.NumpyList(c)
self.assertIsInstance(n, en.NumpyList)
self.assertEqual(len(n), len(x),
f'Length not correct {len(n)}/{len(x)}')
self.assertEqual(len(n.shapes[0]), 1,
f'Shape should only have 1 dim {len(n.shapes[0])}')
self.assertEqual(n.shapes[0][0], len(x),
f'Shape of dim 0 incorrect {n.shapes[0][0]}')
self.assertEqual(len(n.shapes[1]), 1,
f'Shape should only have 1 dim {len(n.shapes[1])}')
self.assertEqual(n.shapes[1][0], len(y),
f'Shape of dim 0 incorrect {n.shapes[1][0]}')
self.assertEqual(n.number_of_lists, len(c),
f'Number of lists incorrect {n.number_of_lists}')
self.assertEqual(n.dtype_names[0], x.dtype.name,
f'dtype not expected {n.dtype_names[0]}')
self.assertEqual(n.dtype_names[1], y.dtype.name,
f'dtype not expected {n.dtype_names[1]}')
self.assertListEqual(
n.lists, c, f'Not the expected return from numpy_list {n.lists}')
def test_creation_wrong_size(self):
x = np.random.rand(5, 2)
y = np.random.rand(2, 2)
with self.assertRaises(en.NumpyListException):
en.NumpyList([x, y])
