def test_prune_rule_1(self):
df = pd.read_csv('tests/test_files/testfile_3.csv',
encoding='utf-8',
delimiter=';')
l = Literal('Sex', 'in', 'Female')
l2 = Literal('Sex', 'in', 'Male')
rule = Rule()
rule.add_literal(l)
rule.add_literal(l2)
ds = DictDataset(1, df)
rule = ds.prune_rule(rule)
self.assertEqual("Sex in Female", rule.to_string())
def test_prune_rule_3(self):
df = pd.read_csv('tests/test_files/mushroom.csv',
encoding='utf-8',
delimiter=';')
ds = DictDataset(0, df)
l = Literal('a5', 'in', ['c', 'f', 'm', 'p', 's', 'y'])
l2 = Literal('a3', 'in', 'e')
rule = Rule()
rule.add_literal(l)
rule.add_literal(l2)
rule = ds.prune_rule(rule)
self.assertEqual("a5 in ['c', 'f', 'm', 'p', 's', 'y']",
rule.to_string())
def test_rule_count_p_n1(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('ClassOfSeat', 'in', '1st')
l2 = Literal('Age', '<', 20)
rule = Rule()
rule.add_literal(l)
rule.add_literal(l2)
ds = DictDataset(1, df)
p, n = ds.count_p_n_rule(rule)
self.assertEqual(1, p)
self.assertEqual(1, n)
def test_delete_not_covered_3(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('ClassOfSeat', 'in', '1st')
l2 = Literal('Age', '<', 20)
rule = Rule()
rule.add_literal(l)
rule.add_literal(l2)
ds = DictDataset(1, df)
len_before = ds.length()
ds.delete_not_covered(rule)
len_after = ds.length()
self.assertEqual(len_before - 30, len_after)
def test_unmake_rule_1(self):
df = pd.read_csv('tests/test_files/mushroom.csv',
encoding='utf-8',
delimiter=';')
ds = BitmapDataset(0, df)
rule = Rule()
l = Literal('a1', 'in', 'x')
l2 = Literal('a3', 'in', 'g')
l3 = Literal('a6', 'in', ['a', 'f'])
rule.add_literal(l)
rule.add_literal(l2)
rule.add_literal(l3)
rule = ds.unmake_rule(rule)
self.assertEqual("[[0, 0], [2, 3], [5, 0], [5, 1]]", str(rule))
def check_literal(self, atr_col_name, best_foil, best_l, i, n0, p0, unique_values, op):
literal = Literal(atr_col_name, op, unique_values[i])
p, n = self.count_p_n_literal(literal)
tmp_foil = count_foil_grow(p0, n0, p, n)
if tmp_foil > best_foil:
best_foil = tmp_foil
best_l = copy.deepcopy(literal)
return best_foil, best_l
def make_rule(self, rule):
if rule is None or len(rule) == 0:
return Rule()
rule = sorted(rule, key=lambda x: x[0])
prev_i = -1
new_rule = Rule()
for i, j in rule:
if prev_i != i:
if prev_i != -1:
l = Literal(self.col_names[prev_i], 'in', values)
new_rule.add_literal(l)
values = list()
values.append(self.col_unique_values[i][j])
prev_i = i
l = Literal(self.col_names[i], 'in', sorted(values))
new_rule.add_literal(l)
return new_rule
def test_literal_count_p_n4(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('Sex', 'in', 'Male')
ds = DictDataset(1, df)
p, n = ds.count_p_n_literal(l)
self.assertEqual(8, p)
self.assertEqual(8, n)
def test_literal_count_p_n3(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('Age', '>', 1000)
ds = DictDataset(1, df)
p, n = ds.count_p_n_literal(l)
self.assertEqual(0, p)
self.assertEqual(4, n)
def test_literal_count_p_n2(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('ClassOfSeat', 'in', ['xab', 'asdas'])
ds = DictDataset(1, df)
p, n = ds.count_p_n_literal(l)
self.assertEqual(0, p)
self.assertEqual(0, n)
def test_rule_count_p_n3(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('ClassOfSeat', 'in', '1st')
rule = Rule()
rule.add_literal(l)
ds = BitmapDataset(1, df)
p, n = ds.count_p_n_rule(ds.unmake_rule(rule))
self.assertEqual(4, p)
self.assertEqual(4, n)
def choose_best_literal(self, atr_col_name, best_foil, best_l, df, n0, p0, unique_values, values_to_literal):
for i in range(0, len(unique_values)):
values_to_literal.append(df.at[i, 'value'])
literal = Literal(atr_col_name, 'in', values_to_literal)
p, n = self.count_p_n_literal(literal)
foil = count_foil_grow(p0, n0, p, n)
if foil > best_foil:
best_foil = foil
best_l = copy.deepcopy(literal)
else:
break
return best_foil, best_l
def test_delete_not_covered_3(self):
df = pd.read_csv('tests/test_files/testfile_8.csv',
encoding='utf-8',
delimiter=';')
l = Literal('ClassOfSeat', 'in', '1st')
rule = Rule()
rule.add_literal(l)
ds = BitmapDataset(1, df)
len_before = ds.length()
ds.delete_not_covered(ds.unmake_rule(rule))
len_after = ds.length()
self.assertEqual(len_before - 24, len_after)
def test_rule_count_p_n5(self):
df = pd.read_csv('tests/test_files/testfile_4.csv',
encoding='utf-8',
delimiter=';')
l = Literal('Sex', 'in', 'Female')
rule = Rule()
rule.add_literal(l)
ds = BitmapDataset(1, df)
ds.delete_covered(ds.unmake_rule(rule))
p, n = ds.count_p_n_rule(ds.unmake_rule(rule))
self.assertEqual(0, p)
self.assertEqual(0, n)
def count_p_n_for_every_value_and_sort(self, atr_col_name, unique_values):
p_to_n = list()
for i in range(0, len(unique_values)):
literal = Literal(atr_col_name, 'in', unique_values[i])
p, n = self.count_p_n_literal(literal)
if n == 0:
p_to_n.append(math.inf)
else:
p_to_n.append(p / n)
df = pd.DataFrame({'value': unique_values, 'p_to_n': p_to_n})
df = df.sort_values(by='p_to_n', ascending=False)
df.index = range(len(df))
return df
def test_delete_not_covered_4(self):
df = pd.read_csv('tests/test_files/testfile_4.csv',
encoding='utf-8',
delimiter=';')
l = Literal('Sex', 'in', 'Male')
rule = Rule()
rule.add_literal(l)
ds = BitmapDataset(1, df)
len_before = ds.length()
ds.delete_covered(ds.unmake_rule(rule))
self.assertEqual(36, ds.length())
ds.delete_not_covered(ds.unmake_rule(rule))
len_after = ds.length()
self.assertEqual(len_before - 64, len_after)
def test_value_covered_by_literal_less_true(self):
l = Literal('test', '<', 15)
self.assertEqual(True, l.value_covered_by_literal(14))
def test_value_covered_by_literal_in_false(self):
l = Literal('test', 'in', ['1st', '2nd'])
self.assertEqual(False, l.value_covered_by_literal('3rd'))
def test_value_covered_by_literal_in_true(self):
l = Literal('test', 'in', ['1st', '2nd'])
self.assertEqual(True, l.value_covered_by_literal('1st'))
def test_value_covered_by_literal_more_false(self):
l = Literal('test', '>', -15)
self.assertEqual(False, l.value_covered_by_literal(-15))
def test_value_covered_by_literal_more_true(self):
l = Literal('test', '>', -15)
self.assertEqual(True, l.value_covered_by_literal(-14))
def test_value_covered_by_literal_less_false(self):
l = Literal('test', '<', 15)
self.assertEqual(False, l.value_covered_by_literal(15))