-
Notifications
You must be signed in to change notification settings - Fork 1
/
conditional_test_add_pure_merge_interval_try.py
378 lines (296 loc) · 20 KB
/
conditional_test_add_pure_merge_interval_try.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
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
import numpy as np
import pandas as pd
import datetime
import os
from sklearn import preprocessing
import metric_utils_merge_interval
from sklearn.metrics import roc_curve, auc, average_precision_score, recall_score, precision_score, f1_score
import heapq
import sys
import matplotlib.pyplot as plt
nowtime = datetime.datetime.now().strftime('%Y_%m_%d_%H-%M-%S')
from sklearn import metrics
import percentage as per
# f = open('./a'+nowtime+'.log', 'a')
# sys.stdout = f
# sys.stderr = f
########################################################################## 一些基础函数 ##########################################################################
def interval_generation(sequence, proportion=None):
if proportion == None:
flag0 = 0
flage = 0
interval = []
for i in range(len(sequence)):
if i != len(sequence) - 1: # 如果不等于最后一个元素的话
if sequence[i + 1] - sequence[i] > 1:
flage = i
interval.append((sequence[flag0], sequence[flage]))
flag0 = i + 1
else:
interval.append((sequence[flag0], sequence[i]))
return interval
else:
flag0 = 0
flage = 0
interval = []
for i in range(len(sequence)):
if i != len(sequence) - 1: # 如果不等于最后一个元素的话
if sequence[i + 1] - sequence[i] > 1:
flage = i
add_len = int((sequence[flage] - sequence[flag0] + 1) * proportion)
if sequence[flage] + add_len < sequence[i + 1] - 2:
interval.append((sequence[flag0], sequence[flage] + add_len))
else:
interval.append((sequence[flag0], sequence[i + 1] - 2))
flag0 = i + 1
else:
interval.append((sequence[flag0], sequence[i]))
return interval
def bianchenglabel(some_interval):
label = np.zeros((cum.ground_truth.shape[0],), dtype=np.int)
for i in range(len(some_interval)):
for j in range(some_interval[i][1] - some_interval[i][0] + 1):
label[some_interval[i][0] + j] = 1
return label
def sort_model_by_time(model_path):
models = os.listdir(model_path)
if not models:
return
else:
files = sorted(models, key=lambda x: os.path.getmtime(os.path.join(model_path, x)))
return files
########################################################################## 类定义 ##########################################################################
class analysis:
def __init__(self, train_rcstr_true, train_rcstr_NN, test_rcstr_true, test_rcstr_NN, train_predict_true, train_predict_NN, test_predict_true, test_predict_NN, ground_truth,
start_t,
end_t, input_win, weight_or_not, weight_type, norm_or_not,rcstr_weight,predict_weight, fuse='A',specific=None):
self.train_rcstr_true = train_rcstr_true
self.train_rcstr_NN = train_rcstr_NN
self.test_rcstr_true = test_rcstr_true
self.test_rcstr_NN = test_rcstr_NN
self.train_predict_true = train_predict_true
self.train_predict_NN = train_predict_NN
self.test_predict_true = test_predict_true
self.test_predict_NN = test_predict_NN
self.power = 6 # power改变点
self.ground_truth = ground_truth
if specific != None:
self.specific = specific
self.start_strptime = datetime.datetime.strptime(start_t, '%Y-%m-%d %I:%M:%S %p')
self.end_strptime = datetime.datetime.strptime(end_t, '%Y-%m-%d %I:%M:%S %p')
self.basetime = datetime.datetime.strptime('2017-10-9 06:02:00 PM', '%Y-%m-%d %I:%M:%S %p')
self.input_win = input_win
self.start = int((self.start_strptime - self.basetime).total_seconds() - self.input_win)
end_index = int((self.end_strptime - self.basetime).total_seconds() - self.input_win)
self.weight_or_not = weight_or_not
self.weight_type = weight_type
self.norm_or_not = norm_or_not
if end_index > ground_truth.shape[0]:
self.end = ground_truth.shape[0]
else:
self.end = end_index + 1
if weight_or_not == True:
if weight_type == 1:
if fuse == 'A':
# 如果相加全从predict口进入
wgt_error_out_predict, wgt_error_thre_predict = metric_utils_merge_interval.error_weighting(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
self.wgt_error_out_ewma = wgt_error_out_predict
self.wgt_error_thre_ewma = wgt_error_thre_predict
elif fuse == 'B':
wgt_error_out_rcstr, wgt_error_thre_rcstr = metric_utils_merge_interval.error_weighting(
self.train_rcstr_true, self.train_rcstr_NN, self.test_rcstr_true, self.test_rcstr_NN,
self.power,
self.input_win, norm_or_not)
wgt_error_out_predict, wgt_error_thre_predict = metric_utils_merge_interval.error_weighting(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
self.wgt_error_out_ewma=rcstr_weight*wgt_error_out_rcstr+predict_weight*wgt_error_out_predict
self.wgt_error_thre_ewma=rcstr_weight*wgt_error_thre_rcstr+predict_weight*wgt_error_thre_predict
elif fuse == 'predict':
wgt_error_out_predict, wgt_error_thre_predict = metric_utils_merge_interval.error_weighting(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
self.wgt_error_out_ewma = wgt_error_out_predict
self.wgt_error_thre_ewma = wgt_error_thre_predict
elif fuse == 'reconstruct':
wgt_error_out_rcstr, wgt_error_thre_rcstr = metric_utils_merge_interval.error_weighting(
self.train_rcstr_true, self.train_rcstr_NN, self.test_rcstr_true, self.test_rcstr_NN,
self.power,
self.input_win, norm_or_not)
self.wgt_error_out_ewma = wgt_error_out_rcstr
self.wgt_error_thre_ewma = wgt_error_thre_rcstr
else:
print('赋值错误')
else:
self.wgt_error_out_ewma, self.wgt_error_thre_ewma = metric_utils_merge_interval.error_weighting_second(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
print("<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<self.wgt_error_thre_ewma>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>:")
print(self.wgt_error_thre_ewma)
else:
if fuse == 'A':
# 如果相加全从predict口进入
nwgt_error_out_predict, nwgt_error_thre_predict = metric_utils_merge_interval.error_no_weighting(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
self.nwgt_error_out_ewma = nwgt_error_out_predict
self.nwgt_error_thre_ewma = nwgt_error_thre_predict
elif fuse == 'B':
nwgt_error_out_rcstr, nwgt_error_thre_rcstr = metric_utils_merge_interval.error_no_weighting(
self.train_rcstr_true, self.train_rcstr_NN, self.test_rcstr_true, self.test_rcstr_NN,
self.power,
self.input_win, norm_or_not)
nwgt_error_out_predict, nwgt_error_thre_predict = metric_utils_merge_interval.error_no_weighting(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
self.nwgt_error_out_ewma=rcstr_weight*nwgt_error_out_rcstr+predict_weight*nwgt_error_out_predict
self.nwgt_error_thre_ewma=rcstr_weight*nwgt_error_thre_rcstr+predict_weight*nwgt_error_thre_predict
elif fuse == 'predict':
nwgt_error_out_predict, nwgt_error_thre_predict = metric_utils_merge_interval.error_no_weighting(
self.train_predict_true, self.train_predict_NN, self.test_predict_true, self.test_predict_NN,
self.power,
self.input_win, norm_or_not)
self.nwgt_error_out_ewma = nwgt_error_out_predict
self.nwgt_error_thre_ewma = nwgt_error_thre_predict
elif fuse == 'reconstruct':
nwgt_error_out_rcstr, nwgt_error_thre_rcstr = metric_utils_merge_interval.error_no_weighting(
self.train_rcstr_true, self.train_rcstr_NN, self.test_rcstr_true, self.test_rcstr_NN,
self.power,
self.input_win, norm_or_not)
self.nwgt_error_out_ewma = nwgt_error_out_rcstr
self.nwgt_error_thre_ewma = nwgt_error_thre_rcstr
else:
print('赋值错误')
self.nwgt_threshold = np.percentile(self.nwgt_error_thre_ewma, 99)
# self.wgt_threshold =0.0018016102467660112
self.offset = self.start + input_win
self.start_time = datetime.timedelta(seconds=self.offset) + self.basetime
print(self.start_time)
self.datetime1 = pd.date_range(self.start_time, periods=self.end - self.start,
freq='S') # train_predict_true.shape[0]
def threshold_grid_search_no_plotting(self, weight_or_not, number, lower, upper, consecutive_or_not, fine_or_not, roc_curve, roc_img_path):
if consecutive_or_not == True:
if weight_or_not == True:
percents, thre_x, precision, recall, f1 = metric_utils_merge_interval.threshold_finetuning_for_consecutive1(
self.wgt_error_out_ewma, self.wgt_error_thre_ewma, number, lower, upper, self.ground_truth,
self.input_win, fine_or_not)
thre_x=np.array(thre_x)
f1 = np.array(f1)
precision = np.array(precision)
recall = np.array(recall)
best = np.argwhere(f1 == max(f1))[0][0]
result_dict = {"模型": self.specific + "(加权的event-based的统计)", '阈值':thre_x[best],"最好的F1值": max(f1),
"此时的Precision": precision[best], "此时的Recall": recall[best]}
return result_dict
else:
percents, thre_x, precision, recall, f1 = metric_utils_merge_interval.threshold_finetuning_for_consecutive1(
self.nwgt_error_out_ewma, self.nwgt_error_thre_ewma, number, lower, upper, self.ground_truth,
self.input_win, fine_or_not)
thre_x = np.array(thre_x)
f1 = np.array(f1)
precision = np.array(precision)
recall = np.array(recall)
best = np.argwhere(f1 == max(f1))[0][0]
result_dict = {"模型": self.specific + "(不加权的event-based的统计)",'阈值':thre_x[best], "最好的F1值": max(f1),
"此时的Precision": precision[best], "此时的Recall": recall[best]}
return result_dict
else: # OUTLIER
# windows = list(range(100,350,50))
windows =[200]
print("window range：", windows)
model_dict_for_window = {"model": [], "window":[],'thres': [], "F1": [], "Precision": [], "Recall": [],
"ratio": [], "percentage": [], "auc":[]}
start_a_model_time = datetime.datetime.now()
for window in windows:
print("################## window = {0} ##############".format(window))
start_a_window_model_time = datetime.datetime.now()
if weight_or_not == True: # weighted
percents, thre_x, precision, recall, f1,fpr,tpr = metric_utils_merge_interval.threshold_finetuning_for_outlier1(
self.wgt_error_out_ewma, self.wgt_error_thre_ewma, number, lower, upper, self.ground_truth,
fine_or_not=fine_or_not, window = window)
################################### 针对一个模型进行阈值搜索后得到的所有结果 ####################################
thre_x = np.array(thre_x)
f1 = np.array(f1)
precision = np.array(precision)
recall = np.array(recall) ## TPR(True postive rate) = TP/(TP+FN) = recall
best = np.argwhere(f1 == max(f1))[0][0]
percentage = per.get_percentage(self.wgt_error_thre_ewma, thre_x[best])
##################################### 对该模型画roc图并计算auc的值 #############################################
auc_i = auc(fpr,tpr)
# print("此时，已经完成了一个模型的阈值搜索，并针对每一个阈值，得到了对应的tpr_i和fpr_i，最终得到了数组tpr, fpr，然后利用auc库函数计算得到该模型的auc")
print("该模型的auc： ", auc_i)
if roc_curve==True:
plt.figure()
lw = 2
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title(self.specific)
plt.plot(fpr, tpr, color='darkorange',label='ROC curve (area = %0.2f)' % auc_i)
plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--') # 这是绘制中间的直线
plt.legend(loc="lower right")
path_image = os.path.join(roc_img_path, 'window-{0}'.format(window)+self.specific+".png")
plt.savefig(path_image, dpi=120)
############################################################################################################
my_dict = {"model": self.specific + "(weighted single point)",'thres':thre_x[best], "F1": max(f1), "Precision": precision[best],
"Recall": recall[best],'percentage':percentage, 'auc':auc_i}
else: # unweighted
percents, thre_x, precision, recall, f1 ,fpr,tpr= metric_utils_merge_interval.threshold_finetuning_for_outlier1(
self.nwgt_error_out_ewma, self.nwgt_error_thre_ewma, number, lower, upper,
self.ground_truth,window = window)
thre_x = np.array(thre_x)
f1 = np.array(f1)
precision = np.array(precision)
recall = np.array(recall)
best = np.argwhere(f1 == max(f1))[0][0]
percentage = per.get_percentage(self.nwgt_error_thre_ewma, thre_x[best])
auc_i = auc(fpr,tpr)
print("该模型的auc： ", auc_i)
if roc_curve==True:
plt.figure()
lw = 2
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title(self.specific)
plt.plot(fpr, tpr, color='darkorange',label='ROC curve (area = %0.2f)' % auc_i)
plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--') # 这是绘制中间的直线
plt.legend(loc="lower right")
path_image = os.path.join(roc_img_path, 'window-{0}'.format(window)+self.specific+".png")
plt.savefig(path_image, dpi=120)
my_dict = {"model": self.specific + "(weighted single point)", 'thres': thre_x[best],
"F1": max(f1), "Precision": precision[best],
"Recall": recall[best], 'percentage': percentage,'auc':auc_i}
# save results
model_dict_for_window["model"].append(my_dict["model"])
model_dict_for_window["thres"].append(my_dict["thres"])
model_dict_for_window["F1"].append(my_dict["F1"])
model_dict_for_window["Precision"].append(my_dict["Precision"])
model_dict_for_window["Recall"].append(my_dict["Recall"])
model_dict_for_window["window"].append(window)
model_dict_for_window["ratio"].append(my_dict["thres"] / max(self.wgt_error_thre_ewma if weight_or_not == True else self.nwgt_error_thre_ewma))
model_dict_for_window["percentage"].append(my_dict["percentage"])
model_dict_for_window["auc"].append(my_dict["auc"])
# get the best f1 under the range of window
f1max = max(model_dict_for_window["F1"])
best = np.argwhere(model_dict_for_window["F1"] == f1max)[0][0]
# print all results
for _ in range(len(windows)):
my_dict = {"model": model_dict_for_window["model"][_], "window":model_dict_for_window["window"][_],"thres": model_dict_for_window["thres"][_], "F1": model_dict_for_window["F1"][_],
"Precision": model_dict_for_window["Precision"][_], "Recall": model_dict_for_window["Recall"][_],
"ratio": model_dict_for_window["ratio"][_], "percentage": model_dict_for_window["percentage"][_],"auc": model_dict_for_window["auc"][_]}
print(my_dict)
result_dict = {"model": model_dict_for_window["model"][best], "window":model_dict_for_window["window"][best],"thres": model_dict_for_window["thres"][best], "F1": model_dict_for_window["F1"][best],
"Precision": model_dict_for_window["Precision"][best], "Recall": model_dict_for_window["Recall"][best],
"ratio": model_dict_for_window["ratio"][best], "percentage": model_dict_for_window["percentage"][best], "auc": model_dict_for_window["auc"][best]}
return result_dict