-
Notifications
You must be signed in to change notification settings - Fork 2
/
app.py
228 lines (178 loc) · 9.84 KB
/
app.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
import globals as gb
from Visualize import Visualize
from SignalMerge import SignalMerge
from SignalFeatures import SignalFeatures
from ModeTracking import ModeTracking
from sklearn.metrics import silhouette_score, adjusted_rand_score, adjusted_mutual_info_score, \
homogeneity_completeness_v_measure
from collections import defaultdict
import datetime
import time
import os
import sys
import random
import numpy as np
import math
import pylab as plt
class App:
def __init__(self, sigReaders):
self.sigReaders = sigReaders
sigsTimeValues = [sr.getSignal() for sr in self.sigReaders]
sigsValues = [values for times, values in sigsTimeValues]
self.sigsRanges = [(min(values), max(values), min(np.gradient(values)), max(np.gradient(values))) for values in
sigsValues]
# -----------------------------------------
def build_features_data(self, d_start=gb.D_START_CLUSTERING, d_end=gb.D_END_CLUSTERING):
print("\n --------- build_features_data ...")
DATA = []
AXES_INFO = []
date = d_start
while date < d_end:
sys.stdout.write("\r%s" % "build_features_data --- " + str(date));
sys.stdout.flush()
sigsTimeValues = [sr.getSignal(start=date, end=gb.DURATION) for sr in self.sigReaders]
date += datetime.timedelta(milliseconds=gb.DURATION)
if any([len(values) < gb.MIN_SUBSEQUENCE_LEN for times, values in sigsTimeValues]):
continue
x = SignalFeatures().extractMany([values for times, values in sigsTimeValues],
self.sigsRanges) # Warning: Future calls to extractMany should take the signals in same order
DATA.append(x)
AXES_INFO.append((x[:], sigsTimeValues))
return DATA, AXES_INFO
# -----------------------------------------
''' Predict from the clustering done in the feature space '''
def predict_fsp(self, d_start, d_end):
dico = defaultdict(list)
timedelta = datetime.timedelta(milliseconds=gb.DURATION)
date = d_start
while date < d_end:
sys.stdout.write("\r%s" % "predict_fsp --- " + str(date));
sys.stdout.flush()
sigsNames = [sr.signal_name for sr in self.sigReaders] # FIXME: Out if the loop
if date + timedelta >= d_end: timedelta = d_end - date
sigsTimeValues = [sr.getSignal(start=date, end=date + timedelta, dated=gb.DATED) for sr in self.sigReaders]
date += timedelta
if any([len(values) < gb.MIN_SUBSEQUENCE_LEN for times, values in sigsTimeValues]):
continue
x = SignalFeatures().extractMany([values for times, values in sigsTimeValues], self.sigsRanges)
y = self.clust.predict(x) # get the cluster id (i.e., cluster label)
for signame, (times, values) in list(zip(sigsNames, sigsTimeValues)):
dico[signame + "TIMES"] += times
dico[signame + "VALUES"] += values
dico[signame + "PREDS"] += [y for _ in values]
# ----------------- Merging signals
sigsTimeValues = [(dico[sr.signal_name + "TIMES"], dico[sr.signal_name + "VALUES"]) for sr in self.sigReaders]
sigsTimeValues.append((dico[self.sigReaders[0].signal_name + "TIMES"],
dico[self.sigReaders[0].signal_name + "PREDS"])) # Add labels as an aditional timeseries
if any([len(values) < gb.MIN_SUBSEQUENCE_LEN for times, values in sigsTimeValues]):
return [], [], []
times, axes = SignalMerge.merge(sigsTimeValues, interpolate=False)
labels = [int(y) for y in axes[-1]]
axes = axes[:-1]
return times, axes, labels
# -----------------------------------------
def predict_ssp(self, d_start, d_end, update=False):
sigsTimeValues = [sr.getSignal(start=d_start, end=d_end, dated=gb.DATED) for sr in self.sigReaders]
if any([len(values) < gb.MIN_SUBSEQUENCE_LEN for times, values in sigsTimeValues]):
return [], [], []
times, axes = SignalMerge.merge(sigsTimeValues, interpolate=False)
X = list(zip(*axes))
labels = []
for t, x in enumerate(X):
pred_mode = self.tracker.track(x, update=update)
labels.append(pred_mode)
sys.stdout.write("\r%s" % "predict_ssp \t pred_mode " + str(pred_mode) + "\t progress " + str(
t * 100. / len(X)) + " " + str(times[t]));
sys.stdout.flush()
return times, axes, labels
# -----------------------------------------
def plot_colored_signals(self, times, axes, labels, path, figname):
viz = Visualize()
signame_labels = [viz.colors[y % len(viz.colors)] for y in labels]
if len(axes) < len(self.sigReaders):
return
for isr, sr in enumerate(self.sigReaders):
figurename = path + sr.signal_name + "_" + str(time.time()) + figname
viz.plot([times, axes[isr]], axs_labels=['Time', sr.signal_name], color=signame_labels, fig=figurename)
figurename = path + "_clustering_projection_AllSignals_" + str(time.time()) + figname
viz.plot(axes, color=signame_labels, fig=figurename)
# -----------------------------------------
def init_clust_tracker(self, clust, AXES_INFO, d_start=gb.D_START_INIT_TRACKER, d_end=gb.D_END_INIT_TRACKER):
print("\n --------- init_clust_tracker ...")
self.clust = clust
self.tracker = ModeTracking(type=gb.PROBA_TYPE)
# ------------- Initialize the Transition and Likelihoods based on the clustering result
labels = []
axes = [[] for _ in self.sigReaders]
for x, sigsTimeValues in AXES_INFO:
sub_times, sub_axes = SignalMerge.merge(sigsTimeValues, interpolate=False)
y = self.clust.predict(x)
sub_labels = [y for _ in sub_times]
labels += sub_labels
for i in range(len(axes)):
axes[i] += sub_axes[i]
self.tracker.update_transition(labels)
self.tracker.update_likelihoods(axes, labels)
# -----------------------------------------
def tracking(self, d_start=gb.D_START_TRACKING, d_end=gb.D_END_TRACKING, path=""):
print("\n --------- tracking ...")
times_fsp, axes_fsp, labels_fsp = [], [], []
times_ssp, axes_ssp, labels_ssp = [], [], []
timedelta = datetime.timedelta(
milliseconds=60 * 60 * 1000) # read chunk by chunk (each chunk is of 'timedelta' milliseconds)
date = d_start
while date < d_end:
if date + timedelta >= d_end: timedelta = d_end - date
times, axes, labels = self.predict_fsp(d_start=date, d_end=date + timedelta)
# self.plot_colored_signals(times, axes, labels, path, figname="_FSP.png")
times_fsp += times;
axes_fsp += axes;
labels_fsp += labels
times, axes, labels = self.predict_ssp(d_start=date, d_end=date + timedelta, update=True)
# self.plot_colored_signals(times, axes, labels, path, figname="_SSP.png")
times_ssp += times;
axes_ssp += axes;
labels_ssp += labels
date += timedelta
# ----------------------------
if gb.ARTIFICIAL:
times, values, true_labels = self.sigReaders[0].getSignal(start=d_start, end=d_end, dated=gb.DATED,
get_modes=True)
ari_fps = adjusted_rand_score(true_labels, labels_fsp);
ari_sps = adjusted_rand_score(true_labels, labels_ssp)
ami_fps = adjusted_mutual_info_score(true_labels, labels_fsp);
ami_sps = adjusted_mutual_info_score(true_labels, labels_ssp)
ho_fps, com_fps, vm_fps = homogeneity_completeness_v_measure(true_labels, labels_fsp);
ho_sps, com_sps, vm_sps = homogeneity_completeness_v_measure(true_labels, labels_ssp)
print("---------------------------------------------------")
print("adjusted_rand_score \t (ari_fps, ari_sps)", (ari_fps, ari_sps))
print("adjusted_mutual_info \t (ami_fps, ami_sps)", (ami_fps, ami_sps))
print("homogeneity \t (ho_fps, ho_sps)", (ho_fps, ho_sps))
print("completeness \t (com_fps, com_sps)", (com_fps, com_sps))
print("v_measure \t (vm_fps, vm_sps)", (vm_fps, vm_sps))
#return (ari_fps, ari_sps), (ami_fps, ami_sps), (ho_fps, ho_sps), (com_fps, com_sps), (vm_fps, vm_sps)
return ((ari_fps, ari_sps), (ami_fps, ami_sps), (ho_fps, ho_sps), (com_fps, com_sps), (vm_fps, vm_sps)), (times_fsp,axes_fsp,labels_fsp,times_ssp,axes_ssp,labels_ssp)
else:
return 0., 0.
# return self.silhouette(axes_fsp, labels_fsp), self.silhouette(axes_ssp, labels_ssp)
# -----------------------------------------
def silhouette(self, axes, labels):
limit = 10000 # FIXME: this is a quick hack to avoid memory errors (for big amount of data)
X = list(zip(*axes))
indexs = range(len(X))
random.shuffle(indexs)
X = np.array([X[i] for i in indexs[:limit]])
Y = np.array([labels[i] for i in indexs[:limit]])
return silhouette_score(X, Y, metric='euclidean')
# -----------------------------------------
def logInformations(self, id_combin, clust, path=""):
print("\n id_combin", id_combin, "k", clust.k)
log = open(os.path.split(path)[0] + '/combins.txt', 'a')
log.write("COMB " + str(id_combin) + '\n')
if clust.ids is not None:
log.write(' - '.join(str(id) for id in clust.ids) + '\n')
log.write(' - '.join(SignalFeatures().getFeaturesName(clust.ids)) + '\n')
log.write('\n')
log.close()
clust.plot(fig=path + '_clustering.png')
# =================================================================