def testPandasIntToMatrix(self):
"""
Test that a matrix holding ints is properly turned into a double matrix.
"""
d = pd.DataFrame({'a': range(5)})
m, _ = to_matrix(d)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.shape[0], 5)
self.assertEqual(m.shape[1], 1)
for i in range(5):
self.assertEqual(m[i], i)
def testNumpyToMatrix(self):
"""
Make sure we can convert a numpy matrix without copying anything.
"""
m1 = np.random.randn(100, 5)
m2, _ = to_matrix(m1)
self.assertTrue(isinstance(m2, np.ndarray))
self.assertEqual(m2.dtype, np.dtype(np.double))
p1 = m1.__array_interface__
p2 = m2.__array_interface__
self.assertEqual(p1['data'], p2['data'])
def testPandasToMatrix(self):
"""
Test that a simple pandas numeric matrix can be turned into a numpy ndarray.
"""
d = pd.DataFrame(np.random.randn(100, 4), columns=list('abcd'))
m, _ = to_matrix(d)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.shape[0], 100)
self.assertEqual(m.shape[1], 4)
self.assertEqual(m.dtype, np.dtype(np.double))
colnames = list('abcd')
for i in range(m.shape[1]):
for j in range(m.shape[0]):
self.assertEqual(m[j, i], d[colnames[i]][j])
def testMultitypeArraylikeToMatrix(self):
"""
Test that if we pass an array with multiple types, we get back the right
thing. The numpy ndarray should be filled with doubles only.
"""
a = [[0.01, 0.02, 3], [0.04, 0.05, 6], [0.07, 0.08, 9],
[0.10, 0.11, 12]]
m, _ = to_matrix(a)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.dtype, np.dtype(np.double))
self.assertEqual(m.shape[0], 4)
self.assertEqual(m.shape[1], 3)
for i in range(4):
for j in range(3):
self.assertEqual(a[i][j], m[i, j])
def testArraylikeToMatrix(self):
"""
Test that if we pass some array, we get back the right thing. This array
will be filled with doubles only.
"""
a = [[0.01, 0.02, 0.03], [0.04, 0.05, 0.06], [0.07, 0.08, 0.09],
[0.10, 0.11, 0.12]]
m, _ = to_matrix(a)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.dtype, np.dtype(np.double))
self.assertEqual(m.shape[0], 4)
self.assertEqual(m.shape[1], 3)
for i in range(4):
for j in range(3):
self.assertEqual(a[i][j], m[i, j])
def testMultitypeArraylikeToMatrix(self):
"""
Test that if we pass an array with multiple types, we get back the right
thing. The numpy ndarray should be filled with doubles only.
"""
a = [[0.01, 0.02, 3],
[0.04, 0.05, 6],
[0.07, 0.08, 9],
[0.10, 0.11, 12]]
m, _ = to_matrix(a)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.dtype, np.dtype(np.double))
self.assertEqual(m.shape[0], 4)
self.assertEqual(m.shape[1], 3)
for i in range(4):
for j in range(3):
self.assertEqual(a[i][j], m[i, j])
def testArraylikeToMatrix(self):
"""
Test that if we pass some array, we get back the right thing. This array
will be filled with doubles only.
"""
a = [[0.01, 0.02, 0.03],
[0.04, 0.05, 0.06],
[0.07, 0.08, 0.09],
[0.10, 0.11, 0.12]]
m, _ = to_matrix(a)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.dtype, np.dtype(np.double))
self.assertEqual(m.shape[0], 4)
self.assertEqual(m.shape[1], 3)
for i in range(4):
for j in range(3):
self.assertEqual(a[i][j], m[i, j])
def testPandasMixedToMatrix(self):
"""
Test that a matrix with one int and one double feature are transformed
correctly.
"""
d = pd.DataFrame({'a': range(50)})
d['b'] = np.random.randn(50, 1)
self.assertEqual(d['a'].dtype, int)
self.assertEqual(d['b'].dtype, np.dtype(np.double))
m, _ = to_matrix(d)
self.assertTrue(isinstance(m, np.ndarray))
self.assertEqual(m.dtype, np.dtype(np.double))
self.assertEqual(m.shape[0], 50)
self.assertEqual(m.shape[1], 2)
colNames = list('ab')
for i in range(2):
for j in range(50):
self.assertEqual(d[colNames[i]][j], m[j, i])
