-
Notifications
You must be signed in to change notification settings - Fork 5
/
visualization.py
175 lines (149 loc) · 6.95 KB
/
visualization.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
'''
@Describe :
@Author : James Jun
@Date :
'''
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
from sklearn.metrics import roc_curve, auc, precision_recall_curve, mean_squared_error, f1_score, precision_recall_fscore_support,confusion_matrix, precision_score, recall_score, roc_auc_score
def get_topk_scores(total_err_scores, topk=1):
total_err_scores = total_err_scores.T
total_features = total_err_scores.shape[0]
topk_indices = np.argpartition(total_err_scores, range(total_features - topk - 1, total_features), axis=0)[
-topk:]
# 特征最大分数值
total_topk_err_scores = np.sum(np.take_along_axis(total_err_scores, topk_indices, axis=0), axis=0)
return total_topk_err_scores
def get_val_performance_data(total_err_scores, normal_scores, gt_labels, topk=1):
total_err_scores = total_err_scores.T
normal_scores = normal_scores.T
gt_labels = gt_labels.tolist()
total_features = total_err_scores.shape[0]
# 取出最大的topK个scores值的索引值,得到每条样本中分数最大的特征索引值
topk_indices = np.argpartition(total_err_scores, range(total_features - topk - 1, total_features), axis=0)[
-topk:]
total_topk_err_scores = []
topk_err_score_map = []
# 取出最大的topK个scores值
total_topk_err_scores = np.sum(np.take_along_axis(total_err_scores, topk_indices, axis=0), axis=0)
# 得到阈值
thresold = np.max(normal_scores)
# 根据标签得到最终预测的label
pred_labels = np.zeros(len(total_topk_err_scores))
pred_labels[total_topk_err_scores > thresold] = 1
for i in range(len(pred_labels)):
pred_labels[i] = int(pred_labels[i])
gt_labels[i] = int(gt_labels[i])
pre = precision_score(gt_labels, pred_labels)
rec = recall_score(gt_labels, pred_labels)
f1 = f1_score(gt_labels, pred_labels)
C = confusion_matrix(gt_labels, pred_labels)
auc_score = roc_auc_score(gt_labels, total_topk_err_scores)
return {
"f1": f1,
"precision": pre,
"recall": rec,
"TP": C[0, 0],
"TN": C[1, 1],
"FP": C[0, 1],
"FN": C[1, 0],
"threshold": thresold,
"latency": 0,
"roc_auc": auc_score,
"pred_labels": pred_labels,
}
if __name__ == '__main__':
path = "11082021_112415"
test_predicted = np.load(f'./output/SWat/{path}/test_predict.npy')
test_true = np.load(f'./output/SWat/{path}/test_actual.npy')
test_recons = np.load(f'./output/SWat/{path}/test_recons.npy')
test_labels = np.load(f'./output/SWat/{path}/test_label.npy')
train_predicted = np.load(f'./output/SWat/{path}/train_predict.npy')
train_recons = np.load(f'./output/SWat/{path}/train_recons.npy')
train_true = np.load(f'./output/SWat/{path}/train_actual.npy')
train_labels = np.load(f'./output/SWat/{path}/train_label.npy')
val_anomaly_scores = np.load(f'./output/SWat/{path}/val_anomaly_scores.npy')
test_anomaly_scores = np.load(f'./output/SWat/{path}/test_anomaly_scores.npy')
val_total_topk_err_scores = get_topk_scores(val_anomaly_scores)
test_total_topk_err_scores = get_topk_scores(test_anomaly_scores)
m_eval = get_val_performance_data(test_anomaly_scores, val_anomaly_scores, test_labels, topk=1)
pred_labels = m_eval["pred_labels"]
# test_scores = np.load('./output/SWat/03082021_194902/test_anomaly_scores.npy')
# total_topk_err_scores = np.load('./output/SWat/03082021_194902/test_score.npy')
# for i in range(len(test_true.columns)):
# plt.figure()
# plt.scatter(np.arange(len(train_true)), train_true[:,i], c='red', label='train_true', s=2)
# plt.scatter(np.arange(len(train_predicted)), train_predicted[:,i], c='blue', label='train_predicted', s=2)
# plt.legend(loc='upper left', fontsize=8)
# plt.show()
# for i in range(train_true.shape[1]):
# plt.figure()
# plt.scatter(np.arange(len(train_true)), train_true[:,i], c='red', label='train_true', s=2)
# plt.scatter(np.arange(len(train_predicted)), train_predicted[:,i], c='blue', label='train_predicted', s=2)
# plt.scatter(np.arange(len(train_recons)), train_recons[:,i], c='green', label='train_recons', s=2)
# plt.legend(loc='upper left', fontsize=8)
# plt.show()
for i in range(test_true.shape[1]):
plt.figure()
plt.scatter(np.arange(len(test_true)), test_true[:,i], c='red', label='test_true', s=2)
plt.scatter(np.arange(len(test_predicted)), test_predicted[:,i], c=test_labels, label='test_predicted', s=2)
plt.scatter(np.arange(len(test_recons)), test_recons[:,i], c=test_labels, label='test_recons', s=2)
plt.legend(loc='upper left', fontsize=8)
plt.show()
plt.figure()
plt.scatter(np.arange(len(test_total_topk_err_scores)), np.atleast_2d(test_total_topk_err_scores).T, c=test_labels, s=2)
plt.legend(['test_scores'])
plt.show()
plt.figure()
plt.scatter(np.arange(len(test_total_topk_err_scores)), np.atleast_2d(test_total_topk_err_scores).T, c=pred_labels, s=2)
plt.legend(['test_scores'])
plt.show()
# plt.figure()
# plt.scatter(np.arange(len(total_topk_err_scores)), total_topk_err_scores, c=test_labels, s=2)
# plt.legend(['total_topk_err_scores'])
# plt.show()
#
# total_topk_err_scores_ = get_topk_scores(test_scores, 2)
# plt.figure()
# plt.scatter(np.arange(len(total_topk_err_scores_)), total_topk_err_scores_, c=test_labels, s=2)
# plt.legend(['total_topk_err_scores_'])
# plt.show()
# plt.figure(0)
# plt.subplot(411)
# plt.scatter(np.arange(len(test_predicted)), test_predicted[:,0], c=test_labels[:,0], s=2)
#
# plt.subplot(412)
# plt.scatter(np.arange(len(test_true)), test_true[:,0], c=test_labels[:,0], s=2)
#
# plt.subplot(413)
# plt.scatter(np.arange(len(total_topk_err_scores)), total_topk_err_scores, c=test_labels[:,0], s=2)
#
# plt.subplot(414)
# plt.scatter(np.arange(len(test_scores)), test_scores[:,0], c=test_labels[:,0], s=2)
#
# plt.show()
#
# train_orig = pd.read_csv(f'./data/SWat/train.csv', sep=',', header=1)
# test_orig = pd.read_csv(f'./data/SWat/test.csv', sep=',')
#
# train_orig[' Timestamp'] = pd.to_datetime(train_orig[' Timestamp'])
# th_time = train_orig[' Timestamp'][0] + pd.Timedelta('6H') # 两个datetime值之间的差(如日,秒和微妙)
# train_orig = train_orig[train_orig[' Timestamp'] > th_time]
#
# train, test = train_orig.iloc[:100000, :], test_orig.iloc[:50000, :]
#
# plt.figure(1)
# plt.subplot(311)
# plt.scatter(np.arange(len(test)), test[:,0], c=test[:,-1], s=2)
#
# plt.subplot(312)
# plt.scatter(np.arange(len(test)), test[:,1], c=test_labels[:,-1], s=2)
#
# plt.subplot(313)
# plt.scatter(np.arange(len(test)), test[:,2], c=test_labels[:,-1], s=2)
#
# plt.show()