def test_default_rule_correct(self):
cba = CBA(support=0.9)
cba_m2 = CBA(support=0.9)
header1 = ["A", "B", "Y"]
rows1 = [
[1, 1, 0],
[0, 0, 1],
]
transactions = TransactionDB(rows1, header1)
cba.fit(transactions)
cba_m2.fit(transactions)
default_class = cba.clf.default_class
default_class_m2 = cba_m2.clf.default_class
self.assertTrue(default_class in ["0", "1"])
self.assertTrue(default_class_m2 in ["0", "1"])
default_class_support = cba.clf.default_class_support
default_class_confidence = cba.clf.default_class_confidence
default_class_support_m2 = cba_m2.clf.default_class_support
default_class_confidence_m2 = cba_m2.clf.default_class_confidence
self.assertTrue(0 <= default_class_support <= 1)
self.assertTrue(0 <= default_class_support_m2 <= 1)
self.assertTrue(0 <= default_class_confidence <= 1)
self.assertTrue(0 <= default_class_confidence_m2 <= 1)
def fmax(param_dict):
print(param_dict)
support, confidence = param_dict["support"] / 1000, param_dict[
"confidence"] / 1000
print(dict(support=support, confidence=confidence))
cba = CBA(support=support, confidence=confidence)
cba.fit(txns)
cba_clf = cba.clf
ids = IDS()
ids_clf = IDSClassifier(
IDSRuleSet.from_cba_rules(cba_clf.rules).ruleset)
ids_clf.quant_dataframe_train = quant_df
ids_clf.calculate_default_class()
ids.clf = ids_clf
metrics = ids.score_interpretability_metrics(quant_df_test)
if not is_solution_interpretable(metrics):
distance = solution_interpretability_distance(metrics)
print(-distance)
return -distance
auc = ids.score_auc(quant_df_test)
print(auc)
return auc
def run_cba(Xtr,
Ytr,
Xt,
Yt,
lb,
support=0.20,
confidence=0.5,
k=None,
log=None):
txns_train = TransactionDB.from_DataFrame(pd.concat([Xtr, Ytr], axis=1))
txns_test = TransactionDB.from_DataFrame(pd.concat([Xt, Yt], axis=1))
cba = CBA(support=support, confidence=confidence, algorithm="m1")
cba.fit(txns_train)
if k is not None:
cba.clf.rules = cba.clf.rules[:k]
Y_pred = [int(i) for i in cba.predict(txns_test)]
for r in cba.clf.rules:
r.covered = set(
[i for i, rd in enumerate(txns_train) if r.antecedent <= rd])
if log is None:
from logger import log
log('cba-k', len(cba.clf.rules))
log('cba-rules', str(cba.clf.rules))
[log('cba-nconds', len(r), i) for i, r in enumerate(cba.clf.rules)]
log('cba-auc', roc_auc_score(lb.transform(Yt.values),
lb.transform(Y_pred)))
log('cba-bacc', balanced_accuracy_score(Yt, Y_pred))
log('cba-disp', dispersion_(cba.clf.rules, average=True))
log('cba-overlap', overlap(cba.clf.rules))
print(confusion_matrix(Yt, Y_pred))
def test_fitting(self):
cba = CBA()
test_dataframe = pd.read_csv(dataset_file, sep=",")
transactions = TransactionDB.from_DataFrame(test_dataframe)
cba.fit(transactions)
def test_predict_probability(self):
cba = CBA(algorithm="m2")
test_dataframe = pd.read_csv(dataset_file, sep=",")
transactions = TransactionDB.from_DataFrame(test_dataframe)
transactions_test = TransactionDB.from_DataFrame(test_dataframe[:2])
cba.fit(transactions)
cba.predict_probability(transactions_test)
def test_predict_probablity(self):
header1 = ["A", "B", "Y"]
rows1 = [[1, 1, 0], [1, 1, 0], [1, 1, 1], [0, 0, 0], [0, 0, 1],
[0, 0, 1]]
transactions = TransactionDB(rows1, header1)
cba = CBA()
cba.fit(transactions)
probs = cba.clf.predict_probability_all(transactions)
def test_accuracy(self):
expected_accuracy = 0.5
cba = CBA(algorithm="m2")
test_dataframe = pd.read_csv(dataset_file, sep=",")
transactions = TransactionDB.from_DataFrame(test_dataframe)
transactions_test = TransactionDB.from_DataFrame(test_dataframe[:2])
cba.fit(transactions)
accuracy = cba.rule_model_accuracy(transactions_test)
self.assertAlmostEqual(accuracy, expected_accuracy, places=3)
def test_predict_probability_works(self):
cba = CBA(algorithm="m1")
test_dataframe = pd.read_csv(dataset_file, sep=",")
transactions = TransactionDB.from_DataFrame(test_dataframe)
transactions_test = TransactionDB.from_DataFrame(test_dataframe[:2])
cba.fit(transactions)
probabilities = cba.predict_probability(transactions_test)
matched_rules = cba.predict_matched_rules(transactions_test)
for idx in range(len(probabilities)):
self.assertEqual(probabilities[idx], matched_rules[idx].confidence)
def test_target_class_works(self):
cba = CBA(algorithm="m2")
test_dataframe = pd.read_csv(dataset_file, sep=",")
transactions = TransactionDB.from_DataFrame(test_dataframe,
target="Gender")
cba.fit(transactions)
rules = cba.clf.rules
rule0 = rules[0]
self.assertEqual(rule0.consequent[0], "Gender")
def test_rule_class_label_works(self):
cba = CBA(algorithm="m2")
test_dataframe = pd.read_csv(dataset_file, sep=",")
transactions = TransactionDB.from_DataFrame(test_dataframe)
cba.fit(transactions)
rules = cba.clf.rules
rule0 = rules[0]
self.assertEqual(rule0.consequent[0],
test_dataframe.columns.values[-1])
def test_inspect(self):
cba = CBA()
test_dataframe = pd.read_csv(dataset_file, sep=";")
transactions = TransactionDB.from_DataFrame(test_dataframe)
cba.fit(transactions)
clf = cba.clf
inspect_df = clf.inspect()
self.assertEqual(type(inspect_df), pd.DataFrame)
self.assertEqual(len(inspect_df), len(clf.rules) + 1)
self.assertEqual(inspect_df["lhs"].iloc[-1], "{}")
def misuse(train_df, test_df):
print("\nMisuse Data")
train = train_df.copy()
test = test_df.copy()
train = train[(train['Label'] != 'Benign') == True]
txns_train = TransactionDB.from_DataFrame(train, target="Label")
txns_test = TransactionDB.from_DataFrame(test)
print("Association Rule Generation")
cba = CBA(support=0.01, confidence=0.01)
cba.fit(txns_train)
predict = cba.predict(txns_test)
test['predict'] = predict
return test
def fmax(param_dict):
print(param_dict)
support, confidence = param_dict["support"] / 1000, param_dict["confidence"] / 1000
print(dict(support=support, confidence=confidence))
cba = CBA(support=support, confidence=confidence)
cba.fit(txns)
cba_clf = cba.clf
ids = IDS()
ids_clf = IDSClassifier(IDSRuleSet.from_cba_rules(cba_clf.rules).ruleset)
ids_clf.quant_dataframe_train = quant_df
ids_clf.calculate_default_class()
ids.clf = ids_clf
auc = ids.score_auc(quant_df_test)
print(auc)
return auc
def train_model(data, support, confidence, rule_length):
print('Mining rules...')
headers = data.columns
data = TransactionDB.from_DataFrame(data)
cba = CBA(support=support,
confidence=confidence,
maxlen=rule_length,
algorithm="m1")
cba.fit(data)
accuracy = cba.rule_model_accuracy(data)
print('Determining semantic coherence...')
semantics.precompute_attributes(headers)
for rule in cba.clf.rules:
antecedent = list(rule.antecedent.itemset.keys())
coherence = semantics.get_rule_coherence(antecedent)
mined_rules.append((f'{rule_to_em_rule(rule)}', rule.confidence,
coherence, rule.consequent.value))
print_model()
print(f'Model accuracy: {round(accuracy * 100, 2)} %')
print(f'Model support: {cba.support}')
support=(1, 999),
confidence=(1, 999),
),
ternary_search_precision=5,
max_iterations=1,
extension_precision=-1,
func_args_extension=dict(support=0,
confidence=0))
best_pars = coord_asc.fit()
print("best_pars:", best_pars)
support, confidence = best_pars[0] / 1000, best_pars[1] / 1000
cba = CBA(support=support, confidence=confidence)
cba.fit(txns)
cba_clf = cba.clf
ids = IDS()
ids_clf = IDSClassifier(
IDSRuleSet.from_cba_rules(cba_clf.rules).ruleset)
ids_clf.quant_dataframe_train = quant_df
ids_clf.calculate_default_class()
ids.clf = ids_clf
data = dict(dataset_name=dataset_name,
algorithm="pyARC",
auc=ids.score_auc(quant_df_test, order_type="cba"),
rule_cutoff=rule_cutoff)
import pandas as pd
from pyarc import CBA, TransactionDB
from sklearn.model_selection import train_test_split
print("")
print("Rule Generation")
data = pd.read_csv('total_data.csv') #disc_data/Known_attack_data
print("Data Size : %s" % str(data.shape))
# 데이터 분할
train, test = train_test_split(data, test_size=0.2, random_state=123)
txns_train = TransactionDB.from_DataFrame(train, target="class")
txns_test = TransactionDB.from_DataFrame(test)
print("Association Rule Generation")
cba = CBA(support=0.1)
cba.fit(txns_train)
print(cba.fit(txns_train))
print("\nRULES : ({})".format(len(cba.clf.rules)))
for i in cba.clf.rules:
print(i)
accuracy = cba.rule_model_accuracy(txns_test)
print("")
print(accuracy)
benchmark_list = []
for dataset_filename in dataset_files:
print(dataset_filename)
df_train = pd.read_csv(os.path.join(dataset_path_train, dataset_filename))
df_test = pd.read_csv(os.path.join(dataset_path_test, dataset_filename))
txns_train = TransactionDB.from_DataFrame(df_train)
txns_test = TransactionDB.from_DataFrame(df_test)
quant_df_train = QuantitativeDataFrame(df_train)
quant_df_test = QuantitativeDataFrame(df_test)
cba = CBA(support=0.1, confidence=0.1)
cba.fit(txns_train)
rules = cba.clf.rules
ids_ruleset = IDSRuleSet.from_cba_rules(rules)
ids = IDS()
ids.clf = IDSClassifier(ids_ruleset.ruleset)
ids.clf.default_class = cba.clf.default_class
metrics_dict = ids.score_interpretability_metrics(quant_df_test)
benchmark_dict = dict(dataset_filename=dataset_filename, algorithm="cba")
benchmark_dict.update(metrics_dict)
print(benchmark_dict)