-
Notifications
You must be signed in to change notification settings - Fork 0
/
gan_evaluator.py
204 lines (161 loc) · 9.42 KB
/
gan_evaluator.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
import numpy as np
import pandas as pd
from sklearn.metrics import classification_report, accuracy_score, f1_score, confusion_matrix
from sklearn.svm import SVC
from sklearn.utils import shuffle
from tensorflow.keras.models import load_model
import seaborn as sns
from utils.plotting import plot_n_lineplots, plot_n_heatmaps
from utils.preprocess import filter_by_activity_index, calc_consultant
from utils.windowing import transform_windows_df, windowing_dataframe
class GanEvaluator:
def __init__(self, generator_file, act_id, labels=['standing', 'walking', 'jogging']):
self.act_id = act_id
self.latent_dim = None
self.generator = self.__load_model(generator_file)
self.labels = labels
# Data Parameters
self.x_train, self.y_train = None, None
self.x_train_clf, self.y_train_clf = None, None
self.x_test_clf, self.y_test_clf = None, None
# Data Processing
self.window_size = None
self.step_size = None
self.col_names = None
self.method = None
self.input_cols_eval = None
# eval
self.best_train_f1_score = None
self.orig_train_acc = None
self.orig_train_f1_score = None
self.best_test_f1_score = None
self.orig_test_acc = None
self.orig_test_f1_score = None
def __load_model(self, generator_file):
model = load_model(generator_file)
self.latent_dim = model.input.shape[1]
return model
def init_data(self, train_path, test_path,
window_size=5 * 50,
step_size=int(5 * 50 / 2),
col_names=['userAcceleration.x', 'userAcceleration.y', 'userAcceleration.z', 'userAcceleration.c'],
method='sliding',
input_cols_train=['userAcceleration.x', 'userAcceleration.y', 'userAcceleration.z'],
input_cols_eval=['userAcceleration.c']
):
self.window_size = window_size
self.step_size = step_size
self.col_names = col_names
self.method = 'sliding'
self.input_cols_eval = input_cols_eval
print('Load Data...')
train_df = pd.read_hdf(train_path)
test_df = pd.read_hdf(test_path)
train_windowed_df = windowing_dataframe(train_df, window_size=window_size, step_or_sample_size=step_size,
col_names=col_names, method=method)
test_windowed_df = windowing_dataframe(test_df, window_size=window_size, step_or_sample_size=step_size,
col_names=col_names, method=method)
print('Transform Data...')
self.x_train, self.y_train = transform_windows_df(train_windowed_df, input_cols=input_cols_train,
one_hot_encode=False,
as_channel=False)
x_train_clf, self.y_train_clf = transform_windows_df(train_windowed_df, input_cols=input_cols_eval,
one_hot_encode=False, as_channel=False)
x_test_clf, self.y_test_clf = transform_windows_df(test_windowed_df, input_cols=input_cols_eval,
one_hot_encode=False, as_channel=False)
self.x_train_clf = x_train_clf.reshape((len(x_train_clf), window_size))
self.x_test_clf = x_test_clf.reshape((len(x_test_clf), window_size))
self.x_train_activity, _ = filter_by_activity_index(x=self.x_train, y=self.y_train, activity_idx=self.act_id)
print('Calculate origin performance...')
self.calc_origin_train_test_performance(True)
print('Done!')
def calc_origin_train_test_performance(self, verbose=False):
if (self.x_train_clf is None) or (self.x_test_clf is None):
print('Please run method: init_data first.')
return
orig_svm_clf = SVC()
orig_svm_clf.fit(self.x_train_clf, self.y_train_clf)
y_train_head = orig_svm_clf.predict(self.x_train_clf)
self.orig_train_acc = accuracy_score(self.y_train_clf, y_train_head)
self.orig_train_f1_score = f1_score(self.y_train_clf, y_train_head, average=None)[self.act_id]
self.best_train_f1_score = self.orig_train_f1_score
y_test_head = orig_svm_clf.predict(self.x_test_clf)
self.orig_test_acc = accuracy_score(self.y_test_clf, y_test_head)
self.orig_test_f1_score = f1_score(self.y_test_clf, y_test_head, average=None)[self.act_id]
self.best_test_f1_score = self.orig_test_f1_score
if verbose:
print('Original training acc: ', self.orig_train_acc)
print('Original training f1_score for act_id ', self.act_id, ': ', self.orig_train_f1_score, '\n')
print('Original test acc: ', self.orig_test_acc)
print('Original test f1_score for act_id ', self.act_id, ': ', self.orig_test_f1_score, '\n')
def generate_data(self, percentage=0.1):
num_gen = int(np.ceil(len(self.x_train_activity) * percentage))
random_latent_vectors = np.random.normal(size=(num_gen, self.latent_dim))
self.generated_sensor_data = self.generator.predict(random_latent_vectors)
gen_df = pd.DataFrame(np.array([ts.transpose() for ts in self.generated_sensor_data]).tolist(),
columns=['userAcceleration.x', 'userAcceleration.y', 'userAcceleration.z'])
gen_df['userAcceleration.c'] = calc_consultant(gen_df)
gen_df['act'] = self.act_id
gen_windowed_df = windowing_dataframe(gen_df, window_size=self.window_size, step_or_sample_size=self.step_size,
col_names=self.col_names, method=self.method)
input_cols = ['userAcceleration.c']
x_gen, y_gen = transform_windows_df(gen_windowed_df, input_cols=input_cols, one_hot_encode=False,
as_channel=False)
x_gen = x_gen.reshape((len(x_gen), self.window_size))
x_train_gen = np.concatenate([self.x_train_clf, x_gen[:num_gen]])
y_train_gen = np.concatenate([self.y_train_clf, y_gen[:num_gen]])
return x_train_gen, y_train_gen, num_gen
def eval_performance(self, x_train_gen, y_train_gen, verbose=False):
gen_svm_clf = SVC()
gen_svm_clf.fit(x_train_gen, y_train_gen)
y_train_head = gen_svm_clf.predict(self.x_train_clf)
gen_train_acc = accuracy_score(self.y_train_clf, y_train_head)
gen_train_f1_score = f1_score(self.y_train_clf, y_train_head, average=None)[self.act_id]
y_test_head = gen_svm_clf.predict(self.x_test_clf)
gen_test_acc = accuracy_score(self.y_test_clf, y_test_head)
gen_test_f1_score = f1_score(self.y_test_clf, y_test_head, average=None)[self.act_id]
if gen_train_f1_score > self.best_train_f1_score or gen_test_f1_score > self.best_test_f1_score:
self.gen_svm_clf = gen_svm_clf
if verbose:
print('Train Acc:', self.orig_train_acc < gen_train_acc, self.orig_train_f1_score < gen_train_f1_score)
print('Origin: ', self.orig_train_acc, ' vs. ', gen_train_acc)
print('F1-Origin', self.orig_train_f1_score, ' vs. ', gen_train_f1_score)
print('\n')
print('Test Performance', self.orig_test_acc < gen_test_acc, self.orig_test_f1_score < gen_test_f1_score)
print('Origin: ', self.orig_test_acc, ' vs. ', gen_test_acc)
print('F1-Origin', self.orig_test_f1_score, ' vs. ', gen_test_f1_score)
print('\n')
print('\n')
return gen_train_f1_score, gen_test_f1_score
def plot_line_plot(self, num=10, random_state=None):
if self.generated_sensor_data is None:
print('Run method generate_data first.')
return
x_train_activity = shuffle(self.x_train_activity, random_state=random_state)
generated_sensor_data = shuffle(self.generated_sensor_data, random_state=random_state)
return plot_n_lineplots(x_train_activity, generated_sensor_data, n=num)
def plot_heat_maps(self, num=10, random_state=None):
if self.generated_sensor_data is None:
print('Run method generate_data first.')
return
x_train_activity = shuffle(self.x_train_activity, random_state=random_state)
generated_sensor_data = shuffle(self.generated_sensor_data, random_state=random_state)
return plot_n_heatmaps(x_train_activity, generated_sensor_data, n=num)
def train_classification_report(self):
y_train_head = self.gen_svm_clf.predict(self.x_train_clf)
print(classification_report(self.y_train_clf, y_train_head))
def test_classification_report(self):
y_test_head = self.gen_svm_clf.predict(self.x_test_clf)
print(classification_report(self.y_test_clf, y_test_head))
def train_cm(self):
y_train_head = self.gen_svm_clf.predict(self.x_train_clf)
cm = confusion_matrix(self.y_train_clf, y_train_head)
cm_df = pd.DataFrame(cm, index=self.labels,
columns=self.labels)
return sns.heatmap(cm_df, annot=True, cmap='YlGnBu', fmt='g')
def test_cm(self):
y_test_head = self.gen_svm_clf.predict(self.x_test_clf)
cm = confusion_matrix(self.y_test_clf, y_test_head)
cm_df = pd.DataFrame(cm, index=self.labels,
columns=self.labels)
return sns.heatmap(cm_df, annot=True, cmap='YlGnBu', fmt='g')