def xgb_scale(self, name='scale_xgb'):
"""
Args:
save_path:
log_path:
X_train:
Y_train:
name:
Returns:
"""
save_log_file(self.log_path)
### 加上了Src、Dst特征
logging.info('使用CIDDS-001数据, 用xgboost完成实验分类,'
'使用xgboost中的sample_weight参数进行样本平衡')
model = machine_learnings.do_xgboost_blance_scale(self.X_train, self.Y_train)
# create HDF5 file
model.save(self.model_path + name + 'model.h5')
def Dnn_sample(self, name='sample_Dnn'):
"""
Args:
save_path:
log_path:
X_train:
Y_train:
name:
Returns:
"""
save_log_file(self.log_path)
logging.info('用dnn完成实验分类,'
'使用sample_weight参数进行样本平衡')
model = dnn.do_dnn_1d_sample(self.X_train, self.Y_train, self.result_path, Input_shape=self.shape)
# create HDF5 file
model.save(self.model_path + name + 'model.h5')
def dnn_test(X_test, Y_test, model_path, log_path, name):
"""
Args:
X_test:
Y_test:
save_path:
name:
Returns:
"""
save_log_file(log_path)
model = load_model(model_path + name + 'model.h5')
Y_test = to_categorical(Y_test, num_classes=5)
Y_pred = model.predict(X_test)
do_metrics(Y_test, Y_pred)
attack_types = ['normal', 'attacker', 'victim', 'suspicious', 'unknown']
confusion_matrixs.plot_confusion_matrix(
np.array(metrics.confusion_matrix(Y_test, Y_pred)),
classes=attack_types,
normalize=True,
title='dnn Normalized confusion matrix')
def xgb_none(self, name='none_xgb'):
"""
Args:
save_path:
log_path:
X_train:
Y_train:
name:
Returns:
"""
save_log_file(self.log_path)
### 无样本平衡
logging.info('使用CIDDS-001数据, 用xgboost完成实验分类,'
'无样本平衡, 设置了划分样本的random_state=1, 并且保存模型')
model = machine_learnings.do_xgboost(self.X_train,self.Y_train)
# create HDF5 file
model.save(self.model_path + name + 'model.h5')
from sources.utils.logfile import save_log_file
from sources.utils.calculation_metrics import do_metrics
from sources.preprocess.features import *
from sources.models import machine_learnings
import logging
if __name__=="__main__":
log_path = '/home/liyulian/code/CIDDS/repositories'
save_log_file(log_path)
### CIDDS-001
path = '/home/liyulian/code/CIDDS/sources/utils/data_features_001.csv'
### 无样本平衡
logging.info('使用CIDDS-001数据, 用xgboost完成实验分类,'
'无样本平衡, 设置了划分样本的random_state=1, 并且保存模型')
X_train, X_test, Y_train, Y_test = get_features_FPPB(path)
Y_pred, model = machine_learnings.do_xgboost(X_train, X_test, Y_train, Y_test)
do_metrics(Y_test, Y_pred)
import pickle # pickle模块
# 保存Model(注:save文件夹要预先建立，否则会报错)
with open('repositories/model_001.pickle', 'wb') as f:
pickle.dump(model, f)
# ### 无样本平衡
# logging.info('使用CIDDS-001数据, 用xgboost完成实验分类,'
