-
Notifications
You must be signed in to change notification settings - Fork 3
/
utils.py
191 lines (148 loc) · 5.78 KB
/
utils.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
import os
from datetime import datetime
import h5py
import pandas as pd
import numpy as np
from matplotlib import pyplot as plt
from model.trainer import cal_regression_metric
time_format = "t%Y_%m_%d_%H_%M_%S"
def list_all_results(name):
results = []
filepath = os.path.join('result', 'result', name + '.h5')
if not os.path.exists(filepath):
raise FileNotFoundError('File not existed.')
with h5py.File(filepath) as h5file:
ts = list(h5file.keys())
ts = map(lambda x: datetime.strptime(x, time_format), ts)
ts = pd.Series(ts).sort_values()
for t in ts:
result = pd.read_hdf(filepath, key=datetime.strftime(t, time_format))
result['name'] = name
result['t'] = t
results.append(result)
results = pd.DataFrame(results)
results = results.set_index(['name', 't'])
return results
def list_multi_results(files):
results = []
for file in files:
result = get_lastest_result(file)
results.append(result)
results = pd.DataFrame(results)
return results
def get_lastest_result(name):
all_results = list_all_results(name)
return all_results.iloc[-1]
def cal_regression_result(name, label_p=0.0, noise_p=0.0):
np.random.seed(1000)
loaded_pre = np.load(os.path.join('result', 'prediction', name + '.npz'))
pre, label = loaded_pre['pre'], loaded_pre['label']
size = pre.shape[0]
label_s = int(size * label_p)
pre[:label_s] = label[:label_s]
noise_s = int(size * noise_p)
np.random.shuffle(pre[:noise_s])
rmse, mae, mape = cal_regression_metric(label, pre, p=False)
print(f'{name}: rmse %.3f, mae %.3f, mape %.3f' % (rmse, mae, mape * 100))
return rmse, mae, mape
def cal_inference_result(name):
loaded = np.load(os.path.join('result', 'prediction', name + '.npz'))
pre, label = loaded['pre'], loaded['label']
print('Overall:')
cal_regression_metric(label, pre)
pre = pre.reshape(pre.shape[0], 3, -1)
label = label.reshape(label.shape[0], 3, -1)
for c in range(3):
print(f'Channel : {c+1}')
cal_regression_metric(label[:, c], pre[:, c])
def plot_single_axis(x, d, lim, title=None, name='undefined', ylabel=None):
figure = plt.figure(figsize=(3, 1.7), dpi=150)
ax = figure.add_subplot(111)
ax.plot(d, marker='o', color='black')
ax.set_xticks(range(len(d)), x)
ax.set_ylim(lim)
ax.tick_params(axis='both', direction='in')
if ylabel is not None:
ax.set_ylabel(ylabel)
if title is not None:
plt.title(title)
plt.savefig(os.path.join('result', 'plot', name + '.pdf'), format='pdf',
bbox_inches='tight', transparent=True)
plt.close()
def plot_dual_single_axis(x, d1, d2, lim, title=None, name='undefined', ylabel=None, legend=None):
figure = plt.figure(figsize=(3, 1.7), dpi=150)
ax = figure.add_subplot(111)
ax.plot(d1, marker='o', color='#1A237E')
ax.plot(d2, marker='v', color='#BF360C')
ax.set_xticks(range(len(d1)), x)
if legend is not None:
ax.legend(legend, ncols=1)
ax.set_ylim(lim)
ax.tick_params(axis='both', direction='in')
if ylabel is not None:
ax.set_ylabel(ylabel)
if title is not None:
plt.title(title)
plt.savefig(os.path.join('result', 'plot', name + '.pdf'), format='pdf',
bbox_inches='tight', transparent=True)
plt.close()
def plot_double_axis(x, d1, d2, title, lim1=None, lim2=None, name='undefined',
ylabel=['RMSE (minute)', 'MAE (minute)']):
figure = plt.figure(figsize=(3, 1.7), dpi=150)
c1 = '#1A237E'
c2 = '#BF360C'
if lim1 is None:
lim1 = [min(d1) - 1, max(d1) + 0.5]
if lim2 is None:
lim2 = [min(d2) - 0.5, max(d2) + 1]
ax1 = figure.add_subplot(111)
p1 = ax1.plot(d1, color=c1, marker='o')
ax1.set_xticks(range(len(d1)), x)
ax1.set_ylim(lim1)
ax1.tick_params(axis='both', direction='in')
ax1.tick_params(axis='y', colors=c1)
ax1.set_ylabel(ylabel[0], color=c1)
ax2 = ax1.twinx()
p2 = ax2.plot(d2, color=c2, marker='v')
ax2.tick_params(axis='y', colors=c2, direction='in')
ax2.set_ylim(lim2)
ax2.set_ylabel(ylabel[1], color=c2)
# plt.legend(handles=p1+p2, labels=['RMSE', 'MAE'], ncol=2, loc='best')
plt.title(title)
plt.savefig(os.path.join('result', 'plot', name + '.pdf'), format='pdf',
bbox_inches='tight')
plt.close()
def plot_image_compare(img1, img2, name='undefined'):
cmaps = ['GnBu', 'YlGn', 'OrRd']
num_c = img1.shape[0]
figure = plt.figure(figsize=(num_c*4, 2*4), dpi=150)
for c in range(num_c):
ax1 = figure.add_subplot(2, num_c, c + 1)
ax1.invert_yaxis()
ax1.imshow(img1[c], cmap=cmaps[c % 3])
ax1.axis('off')
ax2 = figure.add_subplot(2, num_c, num_c + c + 1)
ax2.invert_yaxis()
ax2.imshow(img2[c], cmap=cmaps[c % 3])
ax2.axis('off')
plt.subplots_adjust(wspace=0.1, hspace=0.1)
plt.savefig(os.path.join('result', 'plot', name + '.pdf'), format='pdf', bbox_inches='tight')
plt.close()
def plot_bar(d, name='undefined'):
colors = ['#9CCC65', '#02fdff', '#3cc3ff', '#807fff', '#c03fff', '#fc02ff', '#ff7323']
bar_width = 0.1
fig, ax = plt.subplots(figsize=(3, 2), dpi=150)
for i, v in enumerate(d):
ax.bar(bar_width * i, v, bar_width, color=colors[i])
plt.xticks([], [])
min_max = [np.min(d), np.max(d)]
span = min_max[1] - min_max[0]
plt.ylim([min_max[0] - span * 0.1, min_max[1] + span * 0.1])
plt.savefig(os.path.join('result', 'plot', f'{name}.pdf'), format='pdf',
bbox_inches='tight', transparent=True)
plt.close()
def prob_model_size(model):
param_size = 0
for param in model.parameters():
param_size += param.nelement() * param.element_size()
return param_size