1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
|
# Test methods with long descriptive names can omit docstrings
# pylint: disable=missing-docstring
import unittest
import numpy as np
import scipy.sparse.linalg as sla
import Orange
from Orange.projection import CUR
from Orange.tests import test_filename
class TestCUR(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.ionosphere = Orange.data.Table(test_filename('datasets/ionosphere.tab'))
def test_cur_projection(self):
self.__projection_test_helper(self.ionosphere, rank=10, max_error=10)
self.__projection_test_helper(self.ionosphere, rank=20, max_error=10)
self.__projection_test_helper(self.ionosphere, rank=5, max_error=1)
def __projection_test_helper(self, data, rank, max_error):
cur = CUR(rank=rank, max_error=max_error)
cur_model = cur(data)
self.assertEqual(data.X.shape[0], cur_model.C_.shape[0])
self.assertEqual(data.X.shape[1], cur_model.R_.shape[1])
np.testing.assert_array_equal(cur_model(data).X, cur_model.C_)
def test_cur_reconstruction(self):
self.__reconstruction_test_helper(self.ionosphere, rank=20, max_error=5)
self.__reconstruction_test_helper(self.ionosphere, rank=25, max_error=1)
self.__reconstruction_test_helper(self.ionosphere, rank=30, max_error=0.1)
def __reconstruction_test_helper(self, data, rank, max_error):
U, s, V = sla.svds(data.X, rank)
S = np.diag(s)
X_k = np.dot(U, np.dot(S, V))
err_svd = np.linalg.norm(data.X - X_k, 'fro')
cur = CUR(rank=rank, max_error=max_error, compute_U=True, random_state=0)
cur_model = cur(data)
X_hat = np.dot(cur_model.C_, np.dot(cur_model.U_, cur_model.R_))
err_cur = np.linalg.norm(data.X - X_hat, 'fro')
self.assertLess(err_cur, (3 + cur_model.max_error) * err_svd)
def test_cur_axis(self):
data1 = self.ionosphere[:100]
data2 = self.ionosphere[100:]
cur = CUR(rank=5, max_error=1)
cur_model = cur(data1)
data2_trans1 = cur_model(data2)
data2_trans2 = data2.X[:, cur_model.features_]
np.testing.assert_array_equal(data2_trans1.X, data2_trans2)
data1_trans1 = cur_model(data1, axis=1)
data1_trans2 = data1.X[cur_model.samples_, :]
np.testing.assert_array_equal(data1_trans1.X, data1_trans2)
|
