def classify(classify_class1=True):
if not os.path.exists(dp.bow_log_dataset_dir):
log_dataset = dp.load_file_save_dataset()
else:
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
if classify_class1:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels1
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels1
label_map = log_dataset.class_label1_map
else:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels2
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels2
label_map = log_dataset.class_label2_map
acc = 0
for i in range(log_dataset.test_num):
class_label = knn.classify(test_features[i], train_features,
train_class_labels, 3)
actual_label = test_class_labels[i]
if class_label == actual_label:
acc += 1
print('正确', class_label)
else:
print('错误, 预测类别：%s, 正确类别：%s' %
(label_map[class_label], label_map[actual_label]))
print('正确率：%.2f%%' % (100.0 * acc / log_dataset.test_num))
def classify(classify_class1=True):
if not os.path.exists(dp.bow_log_dataset_dir):
print('数据预处理并保存中')
log_dataset = dp.load_file_save_dataset()
else:
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
if classify_class1:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels1
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels1
label_map = log_dataset.class_label1_map
class_num = log_dataset.class1_num
else:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels2
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels2
label_map = log_dataset.class_label2_map
class_num = log_dataset.class2_num
train_start_time = time.perf_counter()
softmax_clf = softmax.Softmax(train_features,
train_class_labels,
class_num,
alpha=0.01,
iterations=50)
softmax_clf.train()
train_end_time = time.perf_counter()
predict_start_time = time.perf_counter()
predict_class_labels = softmax_clf.predict(test_features)
acc = 0
for i in range(len(predict_class_labels)):
class_label = predict_class_labels[i]
actual_label = test_class_labels[i]
if class_label == actual_label:
acc += 1
print('正确', label_map[class_label])
else:
print('错误, 预测类别：%s, 正确类别：%s' %
(label_map[class_label], label_map[actual_label]))
print('\n正确率：%.2f%%' % (100.0 * acc / log_dataset.test_num))
predict_end_time = time.perf_counter()
print('Softmax-训练阶段运行时间：%s秒' % (train_end_time - train_start_time))
print('Softmax-预测阶段运行时间：%s秒' % (predict_end_time - predict_start_time))
def classify(classify_class1=True):
if not os.path.exists(dp.bow_log_dataset_dir):
print('数据预处理并保存中')
log_dataset = dp.load_file_save_dataset()
else:
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
test_features = log_dataset.test_features
test_num = log_dataset.test_num
if classify_class1:
train_class_labels = log_dataset.train_class_labels1
test_class_labels = log_dataset.test_class_labels1
class_num = log_dataset.class1_num
label_map = log_dataset.class_label1_map
else:
train_class_labels = log_dataset.train_class_labels2
test_class_labels = log_dataset.test_class_labels2
class_num = log_dataset.class2_num
label_map = log_dataset.class_label2_map
train_start_time = time.perf_counter()
p_words, p_class = nb.train_nb(log_dataset.train_features,
train_class_labels, class_num)
train_end_time = time.perf_counter()
predict_start_time = time.perf_counter()
acc = 0
for i in range(test_num):
class_label = nb.nb_classify(p_words, p_class, test_features[i],
class_num)
if class_label == test_class_labels[i]:
acc += 1
print('正确', label_map[class_label])
else:
print('错误, 预测类别：%s, 正确类别：%s' %
(label_map[class_label], label_map[test_class_labels[i]]))
print('\n正确率：%.2f%%' % (100.0 * acc / test_num))
predict_end_time = time.perf_counter()
print('Naive Bayes-训练阶段运行时间：%s秒' % (train_end_time - train_start_time))
print('Naive Bayes-预测阶段运行时间：%s秒' % (predict_end_time - predict_start_time))
def classify(classify_class1=True):
if not os.path.exists(dp.bow_log_dataset_dir):
print('数据预处理并保存中')
log_dataset = dp.load_file_save_dataset()
else:
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
if classify_class1:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels1
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels1
label_map = log_dataset.class_label1_map
else:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels2
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels2
label_map = log_dataset.class_label2_map
train_start_time = time.perf_counter()
clf = MLPClassifier(hidden_layer_sizes=(100,), solver='adam', activation='relu', random_state=1, verbose=True)
clf.fit(train_features, train_class_labels)
train_end_time = time.perf_counter()
predict_start_time = time.perf_counter()
predict_class_labels = clf.predict(test_features)
acc = 0
for i in range(len(predict_class_labels)):
class_label = predict_class_labels[i]
actual_label = test_class_labels[i]
if class_label == actual_label:
acc += 1
print('正确', label_map[class_label])
else:
print('错误, 预测类别：%s, 正确类别：%s' % (label_map[class_label], label_map[actual_label]))
print('\n正确率：%.2f%%' % (100.0 * acc / log_dataset.test_num))
predict_end_time = time.perf_counter()
print('Neural Network-训练阶段运行时间：%s秒' % (train_end_time - train_start_time))
print('Neural Network-预测阶段运行时间：%s秒' % (predict_end_time - predict_start_time))
def classify(classify_class1=True):
if not os.path.exists(dp.bow_log_dataset_dir):
print('数据预处理并保存中')
log_dataset = dp.load_file_save_dataset()
else:
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
if classify_class1:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels1
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels1
label_map = log_dataset.class_label1_map
else:
test_features = log_dataset.test_features
test_class_labels = log_dataset.test_class_labels2
train_features = log_dataset.train_features
train_class_labels = log_dataset.train_class_labels2
label_map = log_dataset.class_label2_map
train_start_time = time.perf_counter()
knn_classifier = KNeighborsClassifier(n_neighbors=3)
knn_classifier.fit(train_features, train_class_labels)
train_end_time = time.perf_counter()
predict_start_time = time.perf_counter()
acc = 0
for i in range(log_dataset.test_num):
class_label = knn_classifier.predict([test_features[i]])[0]
actual_label = test_class_labels[i]
if class_label == actual_label:
acc += 1
print('正确', label_map[class_label])
else:
print('错误, 预测类别：%s, 正确类别：%s' %
(label_map[class_label], label_map[actual_label]))
print('\n正确率：%.2f%%' % (100.0 * acc / log_dataset.test_num))
predict_end_time = time.perf_counter()
print('KNN-sklearn-训练阶段运行时间：%s秒' % (train_end_time - train_start_time))
print('KNN-sklearn-预测阶段运行时间：%s秒' % (predict_end_time - predict_start_time))
def classify(classify_class1=True):
if not os.path.exists(dp.bow_log_dataset_dir):
print('数据预处理并保存中')
log_dataset = dp.load_file_save_dataset()
else:
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
if classify_class1:
train_class_labels = log_dataset.train_class_labels1
label_map = log_dataset.class_label1_map
test_class_labels = log_dataset.test_class_labels1
else:
train_class_labels = log_dataset.train_class_labels2
label_map = log_dataset.class_label2_map
test_class_labels = log_dataset.test_class_labels2
d_tree = tree.DecisionTreeClassifier()
build_tree_start_time = time.perf_counter()
d_tree.fit(log_dataset.train_features, train_class_labels)
build_tree_end_time = time.perf_counter()
print(tree.export_text(d_tree, feature_names=log_dataset.feature_labels))
predict_start_time = time.perf_counter()
acc = 0
for i in range(log_dataset.test_num):
class_label = d_tree.predict([log_dataset.test_features[i]])[0]
actual_label = test_class_labels[i]
if class_label == actual_label:
acc += 1
print('正确', label_map[class_label])
else:
print('错误, 预测类别：%s, 正确类别：%s' %
(label_map[class_label], label_map[actual_label]))
print('\n正确率：%.2f%%' % (100.0 * acc / log_dataset.test_num))
predict_end_time = time.perf_counter()
print('决策树-sklearn-构建决策树运行时间：%s秒' %
(build_tree_end_time - build_tree_start_time))
print('决策树-sklearn-预测阶段运行时间：%s秒' % (predict_end_time - predict_start_time))
def classify():
log_dataset = dp.load_log_dataset(dp.bow_log_dataset_dir)
train_data = log_dataset.get_combined_train()
d_tree = id3.create_tree(train_data, log_dataset.feature_labels)
print(d_tree)