def test_reset(self):
"""Test if the reset function allows a new algorithm execution."""
bmda = Bmda()
bmda.setup(40)
bmda.cost_func = GbCostFunction(OneMax, var_size=10)
bmda.search()
bmda.reset()
self.assertEqual(bmda.iters, 0)
self.assertIsNotNone(bmda.distr)
self.assertEqual(bmda.cost_func.evals, 0)
def test_max_chisquare_base(self):
chi_matrix = np.array([[0.0, 4.0, 0.0], [4, 0.0, 5.0], [0.0, 5.0,
0.0]])
x, y, chi = Bmda.get_max_dependency([0, 1], [2], chi_matrix)
self.assertEqual(x, 1)
self.assertEqual(y, 2)
self.assertEqual(chi, 5.0)
def test_generate_graph_with_dependencies(self):
"""Test that the algorithm generates a graph with two root nodes"""
pop = np.array(([[0, 0, 1, 0], [1, 1, 0, 0], [1, 1, 0, 0],
[0, 0, 1, 0], [1, 1, 0, 0]]))
roots, _, _ = Bmda.build_graph(pop, np.zeros(4),
DependencyMethod.chi2_test, 3.84)
self.assertEqual(len(roots), 2)
self.assertTrue(np.equal(roots, 3).any())
def test_generate_graph_all_independent(self):
"""Test that the algorithm generates no graph, all variables independent"""
pop = np.array(([[0, 0, 0, 1], [1, 0, 1, 0], [0, 1, 0, 0],
[1, 1, 1, 0], [0, 0, 1, 1]]))
roots, _, _ = Bmda.build_graph(pop, np.zeros(4),
DependencyMethod.chi2_test, 3.84)
roots.sort()
self.assertTrue(np.equal(roots, [0, 1, 2, 3]).all())
def setup_kiedra(data):
train_data, test_data = sample_data(data, train_size)
svm_params = {
'svm_type': orngSVM.SVMLearner.C_SVC,
'kernel_type': orngSVM.SVMLearner.RBF,
'C': 100,
'gamma': 10,
'normalization': False
}
factory = GbFactory(orngSVM.SVMLearner, svm_params)
cost_func = GbWrapperCostFunction(train_data, test_data, factory, 0, 2, 2)
bmda = Bmda()
bmda.setup(pop,
gens,
dependency_method=DependencyMethod.sim,
independence_threshold=0.1)
bmda.cost_func = cost_func
return bmda
def test_shape_calculate_chisquare_matrix(self):
pop = np.array(([[0, 0, 0, 1], [1, 0, 1, 0], [0, 1, 0, 0],
[1, 1, 1, 0], [0, 0, 1, 1], [1, 1, 0, 0]]))
chi_matrix = Bmda.calc_dependency_matrix(pop,
DependencyMethod.chi2_test,
3.84)
self.assertEqual(chi_matrix.shape, (4, 4))
expected = np.array([[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0],
[0, 0, 0, 0]])
#Test for a total independent chi matrix.
chi_matrix = chi_matrix * (np.eye(4, 4) * -1 + 1)
self.assertTrue((chi_matrix == 0.0).all())
def test_basic_search_zero(self):
"""Test class for the Bmda algorithm"""
bmda = Bmda()
bmda.setup(40)
bmda.cost_func = GbCostFunction(ZeroMax, var_size=10)
result = bmda.search()
self.assertTrue((result.params + 1).all())
self.assertGreaterEqual(result.cost, 8.0)
def test_basic_search_onemax(self):
"""Test class for the Bmda algorithm"""
bmda = Bmda()
bmda.setup(10, 40)
bmda.cost_func = GbCostFunction(OneMax)
result = bmda.search()
self.assertGreaterEqual(result.params.sum(), 8.0)
self.assertGreaterEqual(result.cost, 8.0)
def test_basic_search_onemax_mi_method(self):
"""Test class for the Bmda algorithm"""
bmda = Bmda()
bmda.setup(40, dependency_method=DependencyMethod.mi)
bmda.cost_func = GbCostFunction(OneMax, var_size=10)
result = bmda.search()
self.assertGreaterEqual(result.params.sum(), 8.0)
self.assertGreaterEqual(result.cost, 8.0)
def test_basic_search_onemax(self):
"""Test class for the Bmda algorithm"""
var_size = 10
bmda = Bmda()
bmda.setup(40)
bmda.cost_func = GbCostFunction(OneMax, var_size=var_size)
result = bmda.search()
self.assertTrue((result.params + 1).all())
self.assertGreaterEqual(result.cost, var_size * 0.8)