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 test_prune_rule_2(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'])
rule = Rule()
rule.add_literal(l)
rule = ds.prune_rule(rule)
self.assertEqual("a5 in ['c', 'f', 'm', 'p', 's', 'y']",
rule.to_string())
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_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 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_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 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 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 grow_rule(self):
rule = Rule()
growset = DictDataset(self.prod, self.df)
while True:
p0, n0 = growset.count_p_n_rule(rule)
best_foil = -math.inf
best_l = None
for i in range(0, len(list(growset.dict.keys()))):
if self.class_name!=list(growset.dict.keys())[i]:
col_name = list(growset.dict.keys())[i]
foil = best_foil
if col_name not in [x.var_name for x in rule.literals] or col_name in self.numeric_cols:
col_values = list(set(growset.dict[col_name].values()))
l, foil = growset.find_best_literal(p0, n0, col_values, col_name)
if foil > best_foil:
best_l = copy.deepcopy(l)
best_foil = foil
if best_foil > self.grow_param:
rule.add_literal(best_l)
growset.delete_not_covered(rule)
else:
break
return rule