def normalize_feature():
feature_vectors = DBUtil.select_all_features()
feature_vectors = numpy.asarray(feature_vectors)
avg = numpy.mean(feature_vectors)
std = numpy.std(feature_vectors, axis=0)
normalized_vectors = (feature_vectors - avg) / std
normalized_vectors = normalized_vectors.tolist()
mysql_conn = DBUtil.get_mysql_conn()
cur = mysql_conn.cursor()
index = 1
for normalized_vector in normalized_vectors:
if index % 100 == 0:
print index
cur.execute(
"INSERT INTO Normalized_Features (id, visitDuration, bytesTransfer, numLogs2PerIP, numErrorPerIP, numLogsSecure, seriousError, tags, tags2) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)",
(index, normalized_vector[0], normalized_vector[1], normalized_vector[2], normalized_vector[3], normalized_vector[4],
normalized_vector[5], normalized_vector[6], normalized_vector[7]))
mysql_conn.commit()
index = index + 1
cur.close()
mysql_conn.close()
def normalize_feature():
feature_vectors = DBUtil.select_all_features()
size = len(feature_vectors)
feature_vectors = numpy.asarray(feature_vectors)
min = numpy.min(feature_vectors, axis=0)
max = numpy.max(feature_vectors, axis=0)
normalized_vectors = numpy.divide(feature_vectors - min, max - min)
normalized_vectors = normalized_vectors.tolist()
mysql_conn = DBUtil.get_mysql_conn()
cur = mysql_conn.cursor()
index = 1
for normalized_vector in normalized_vectors:
if index % 100 == 0:
print str(index) + " of " + str(size)
# visitDuration, numIP, bytesTransfer, urlLength, numLogs2PerIP, numErrorPerIP, numLogsSecure, unsafeChar
cur.execute(
"INSERT INTO Normalized_Features (id, visitDuration, numIP, bytesTransfer, urlLength, numLogs2PerIP, numErrorPerIP, numLogsSecure, unsafeChar) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)",
(index, normalized_vector[0], normalized_vector[1], normalized_vector[2], normalized_vector[3], normalized_vector[4], normalized_vector[5], normalized_vector[6], normalized_vector[7]))
mysql_conn.commit()
index = index + 1
cur.close()
mysql_conn.close()
