def test_get_filtered_table(self):
A = pd.DataFrame([['a1', 'A', 0.11, 53704]])
A.columns = ['ID', 'name', 'price', 'zip code']
em.set_key(A, 'ID')
B = pd.DataFrame([['b1', 'A', 0.11, 54321]])
B.columns = ['ID', 'name', 'price', 'zip code']
em.set_key(B, 'ID')
A_key = 'ID'
B_key = 'ID'
ltable_col_dict = db._build_col_name_index_dict(A)
rtable_col_dict = db._build_col_name_index_dict(B)
attr_corres = [('ID', 'ID'), ('name', 'name'),
('price', 'price'),
('zip code', 'zip code')]
db._filter_corres_list(A, B, A_key, B_key, ltable_col_dict,
rtable_col_dict, attr_corres)
filtered_A, filtered_B = db._get_filtered_table(A, B, attr_corres)
expected_filtered_A = pd.DataFrame([['a1', 'A']])
expected_filtered_A.columns = ['ID', 'name']
em.set_key(expected_filtered_A, 'ID')
expected_filtered_B = pd.DataFrame([['b1', 'A']])
expected_filtered_B.columns = ['ID', 'name']
em.set_key(expected_filtered_B, 'ID')
self.assertEqual(expected_filtered_A.equals(filtered_A), True)
self.assertEqual(expected_filtered_B.equals(filtered_B), True)
def test_get_filtered_table(self):
A = pd.DataFrame([['a1', 'A', 0.11, 53704]])
A.columns = ['ID', 'name', 'price', 'zip code']
em.set_key(A, 'ID')
B = pd.DataFrame([['b1', 'A', 0.11, 54321]])
B.columns = ['ID', 'name', 'price', 'zip code']
em.set_key(B, 'ID')
A_key = 'ID'
B_key = 'ID'
ltable_col_dict = db._build_col_name_index_dict(A)
rtable_col_dict = db._build_col_name_index_dict(B)
attr_corres = [('ID', 'ID'), ('name', 'name'),
('price', 'price'),
('zip code', 'zip code')]
db._filter_corres_list(A, B, A_key, B_key, ltable_col_dict,
rtable_col_dict, attr_corres)
filtered_A, filtered_B = db._get_filtered_table(A, B, attr_corres)
expected_filtered_A = pd.DataFrame([['a1', 'A']])
expected_filtered_A.columns = ['ID', 'name']
em.set_key(expected_filtered_A, 'ID')
expected_filtered_B = pd.DataFrame([['b1', 'A']])
expected_filtered_B.columns = ['ID', 'name']
em.set_key(expected_filtered_B, 'ID')
self.assertEqual(expected_filtered_A.equals(filtered_A), True)
self.assertEqual(expected_filtered_B.equals(filtered_B), True)
def test_select_features_3(self):
A = read_csv_metadata(path_a, key='ID')
B = read_csv_metadata(path_b, key='ID')
A_key = em.get_key(A)
B_key = em.get_key(B)
corres_list = [(0, 0)]
A_filtered, B_filtered = db._get_filtered_table(A, B, A_key, B_key, corres_list)
actual_selected_features = db._select_features(
A_filtered, B_filtered, A_key)
expected_selected_features = []
self.assertEqual(actual_selected_features, expected_selected_features)
def test_select_features_3(self):
A = read_csv_metadata(path_a, key='ID')
B = read_csv_metadata(path_b, key='ID')
A_key = em.get_key(A)
B_key = em.get_key(B)
cols_A = list(A.columns)
cols_B = list(B.columns)
corres_list = [(cols_A[0], cols_B[0])]
A_filtered, B_filtered = db._get_filtered_table(A, B, corres_list)
actual_selected_features = db._select_features(
A_filtered, B_filtered, A_key, B_key)
expected_selected_features = []
self.assertEqual(actual_selected_features, expected_selected_features)
def test_select_features_2(self):
A = read_csv_metadata(path_a, key='ID')
B = read_csv_metadata(path_b, key='ID')
A_key = em.get_key(A)
B_key = em.get_key(B)
cols_A = list(A.columns)
cols_B = list(B.columns)
corres_list = [(cols_A[0], cols_B[0]), (cols_A[1], cols_B[1]), (cols_A[4],
cols_B[4])]
A_filtered, B_filtered = db._get_filtered_table(A, B, corres_list)
actual_selected_features = db._select_features(
A_filtered, B_filtered, A_key, B_key)
expected_selected_features = [1, 2]
self.assertEqual(actual_selected_features, expected_selected_features)
def test_get_feature_weight_2(self):
A = read_csv_metadata(path_a, key='ID')
B = read_csv_metadata(path_b, key='ID')
A_key = em.get_key(A)
B_key = em.get_key(B)
corres_list = [(0, 0), (1, 1), (4, 4), (5, 5)]
A_filtered, B_filtered = db._get_filtered_table(
A, B, A_key, B_key, corres_list)
A_wlist = db._get_feature_weight(A_filtered)
expected_A_wlist = [2.0, 2.0, 2.0, 1.4]
self.assertEqual(A_wlist, expected_A_wlist)
B_wlist = db._get_feature_weight(B_filtered)
expected_B_wlist = [2.0, 2.0, 2.0, 1.3333333333333333]
self.assertEqual(B_wlist, expected_B_wlist)
def test_get_feature_weight_2(self):
A = read_csv_metadata(path_a, key='ID')
B = read_csv_metadata(path_b, key='ID')
A_key = em.get_key(A)
B_key = em.get_key(B)
cols_A = list(A.columns)
cols_B = list(B.columns)
corres_list = [(cols_A[0], cols_B[0]), (cols_A[1], cols_B[1]),
(cols_A[4], cols_B[4]), (cols_A[5], cols_B[5])]
A_filtered, B_filtered = db._get_filtered_table(
A, B, A_key, B_key, corres_list)
A_wlist = db._get_feature_weight(A_filtered)
expected_A_wlist = [2.0, 2.0, 2.0, 1.4]
self.assertEqual(A_wlist, expected_A_wlist)
B_wlist = db._get_feature_weight(B_filtered)
expected_B_wlist = [2.0, 2.0, 2.0, 1.3333333333333333]
self.assertEqual(B_wlist, expected_B_wlist)
def test_get_feature_weight_2(self):
A = read_csv_metadata(path_a, key='ID')
B = read_csv_metadata(path_b, key='ID')
A_key = em.get_key(A)
B_key = em.get_key(B)
cols_A = list(A.columns)
cols_B = list(B.columns)
corres_list = [(cols_A[0], cols_B[0]), (cols_A[1], cols_B[1]), (cols_A[4],
cols_B[4]),
(cols_A[5], cols_B[5])]
A_filtered, B_filtered = db._get_filtered_table(
A, B, corres_list)
A_wlist = db._get_feature_weight(A_filtered)
expected_A_wlist = [2.0, 2.0, 2.0, 1.4]
self.assertEqual(A_wlist, expected_A_wlist)
B_wlist = db._get_feature_weight(B_filtered)
expected_B_wlist = [2.0, 2.0, 2.0, 1.3333333333333333]
self.assertEqual(B_wlist, expected_B_wlist)
