-
Notifications
You must be signed in to change notification settings - Fork 2
/
Figure_6_b.py
175 lines (134 loc) · 5.72 KB
/
Figure_6_b.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
import prose.datainsights as di
import os
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import subprocess
from sklearn.decomposition import PCA
import numpy as np
import numpy.random as rnd
import matplotlib.pyplot as plt
import os
import sys
import warnings
from matplotlib import rc
from matplotlib import rcParams
import copy
np.random.seed(0)
rcParams['font.family'] = 'sans'
rc('text', usetex=True)
warnings.filterwarnings('ignore')
df_dict = None
data_source = "data/uncompressed/har/"
sensors = ["acc", "Gyroscope"]
activities = ["lying", "running", "sitting", "standing", "walking"]
positions = ["head", "shin", "thigh", "upperarm", "waist", "chest"]
persons = ['person' + str(i) for i in range(1, 16)]
def generate_df_dict():
global df_dict
df_dict = dict()
for person in persons:
if person not in df_dict:
df_dict[person] = dict()
cur_person_data = data_source + person + "/"
for activity in activities:
df = None
for position in positions:
for sensor in sensors:
for file in os.listdir(cur_person_data):
if not file.startswith("_"): continue
if activity in file and sensor in file and position in file:
cur_df = pd.read_csv(cur_person_data + file, index_col=0)
if df is None:
df = cur_df
else:
df = pd.concat([df, cur_df], axis=1, ignore_index=True)
df_dict[person][activity] = pd.DataFrame(df, dtype=float)
generate_df_dict()
def get_data_for_activities(cur_activities, N=100):
if df_dict is None:
print("Re-computing df-dict")
generate_df_dict()
train_df = None
for activity in cur_activities:
for person in persons[4:]:
cur_df = copy.deepcopy(df_dict[person][activity])
cur_df["person"] = person
if train_df is None:
train_df = cur_df
else:
train_df = pd.concat([train_df, cur_df], axis=0)
X = np.array(np.array(train_df)[:,:-1], dtype=float)
X = np.array(np.array(train_df)[:,:-1], dtype=float)
y = np.array(train_df)[:,-1]
sampled_indexes = np.random.choice(range(X.shape[0]), N)
return pd.DataFrame(X[sampled_indexes], dtype=float), X[sampled_indexes], y[sampled_indexes]
sedentary_activities = ['lying', 'sitting', 'standing']
mobile_activities = ['running', 'walking']
import pandas as pd
import numpy as np
import prose.datainsights as di
import matplotlib.pyplot as plt
from sklearn.linear_model import LogisticRegression
from scipy.stats.stats import pearsonr
N = 5000 # number of data points in the train/test set
n_test_cases = 10 # number of experiments to run
step = 5
np.random.seed(1)
score_drops = []
our_violations = []
kk = 10
for kk in range(0, 51, step):
mix_fraction = (kk * N) // 100
print("Fraction of mobile data: ", kk, '%')
score_drop_ = 0
our_violation_ = 0
cs = 0
train_df, train_X, train_y = get_data_for_activities(mobile_activities, N=N)
for _ in range(n_test_cases):
test_df_1, test_X_1, test_y_1 = get_data_for_activities(mobile_activities, N=mix_fraction)
test_df_2, test_X_2, test_y_2 = get_data_for_activities(sedentary_activities, N=N - mix_fraction)
test_df = pd.concat([test_df_1, test_df_2], axis=0)
test_X = np.vstack((test_X_1, test_X_2))
test_y = np.concatenate([test_y_1, test_y_2], axis=0)
assertions = di.learn_assertions(pd.DataFrame(test_X), max_self_violation=0.85)
clf = LogisticRegression(random_state=0, solver='lbfgs', multi_class='multinomial')
clf.fit(test_X, test_y)
max_score = clf.score(test_X, test_y)
score = clf.score(train_X, train_y)
score_drop = max_score - score
our_violation = assertions.evaluate(pd.DataFrame(train_X)).avg_violation
score_drop_ += score_drop
our_violation_ += our_violation
cs += 1
# print(cs, end='')
# print('')
score_drop = score_drop_ / cs
our_violation = our_violation_ / cs
# print("Drift fraction:", mix_fraction / float(N))
# print("ML score drop:", score_drop)
# print("Violations Our:", our_violation)
score_drops.append(score_drop)
our_violations.append(our_violation)
# if len(score_drops) > 1 and cs % 10 == 0:
# print("Done", cs, "tests")
# print("PCC and P-Value", pearsonr(score_drops, our_violations))
systemName = 'CCSynth'
# print(score_drops)
# print(our_violations)
print("PCC & P-Value", pearsonr(score_drops,our_violations))
# our_violations = [0.38666818562854066, 0.1265843622117037, 0.06980683859973677, 0.04284904880355313, 0.02795662730789662, 0.02108232003713504, 0.015215993190774876, 0.011677844742624718, 0.008839468644942597, 0.0063424241999867955, 0.005461418127516672]
# score_drops = [0.7059799999999999, 0.5723, 0.50674, 0.45854, 0.4145199999999999, 0.35785999999999996, 0.3154, 0.27490000000000003, 0.24375999999999998, 0.18882, 0.15572]
fig = plt.gcf()
fig.set_size_inches(3, 3)
rcParams['font.family'] = 'serif'
rc('text', usetex=True)
rcParams['figure.dpi'] = 300
plt.plot(np.arange(1, 12, 1) * 5.0, our_violations, "C3s", label='CCSynth')
plt.plot(np.arange(1, 12, 1) * 5.0, score_drops, "C0o", label="Classifier (LR)")
plt.legend(ncol=1, loc='upper right', handletextpad=0.1, handlelength=1)
plt.ylim([0,1])
plt.xlabel("Noise (\%) during training")
plt.ylabel("CC Violation/acc-drop", fontsize=14)
plt.xticks(np.arange(1, 12, 2) * 5)
plt.savefig(os.path.join("Plots", "Figure_6_b.pdf"), bbox_inches="tight")