forked from nyuad-cai/MedFuse
-
Notifications
You must be signed in to change notification settings - Fork 0
/
plot.py
186 lines (138 loc) · 6.41 KB
/
plot.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
import matplotlib.pyplot as plt
import torch
import glob
import numpy as np
import re
plt.rcParams.update({'font.size': 30})
# plt.rcParams['figure.figsize'] = [8.0, 3.4]
def plot_types_bar():
plt.rcParams.update({'font.size': 18})
keys = ['Acute', 'Mixed', 'Chronic']
single_aurocs = [0.761, 0.749, 0.717]
single_aurocs_errors = np.abs(np.array([(0.732, 0.789), (0.724 , 0.773) , (0.693, 0.741)]).T - single_aurocs)
single_auprcs = [0.432, 0.458, 0.487]
single_auprcs_errors = np.abs(np.array([(0.386, 0.486) , (0.413, 0.506) , (0.448, 0.530)]).T - single_auprcs)
multi_aurocs = [0.772, 0.800, 0.745]
multi_aurocs_errors = np.abs(np.array([(0.744, 0.800) , (0.776 , 0.823) , (0.723, 0.768)]).T - multi_aurocs)
multi_auprcs = [0.433, 0.565, 0.512]
multi_auprcs_errors = np.abs(np.array([(0.386, 0.486) , (0.516, 0.614) , (0.473, 0.555)]).T - multi_auprcs)
barWidth = 0.05
xs1 = (np.arange(len(single_aurocs)) ) * 0.2
xs2 = [x + barWidth for x in xs1]
plt.bar(xs1, single_aurocs, color ='#FF1F5B', width = barWidth,
label ='Uni-modal')
plt.bar(xs2, multi_aurocs, color ='#009ADE', width = barWidth,
label ='Multi-modal')
plt.errorbar(xs1, single_aurocs, color ='black', yerr = single_aurocs_errors, fmt='o')
plt.errorbar(xs2, multi_aurocs, color ='black', yerr = multi_aurocs_errors, fmt='o')
plt.ylabel('AUROC')#, fontweight ='bold', fontsize = 15)
locs = [(r + barWidth/2) for r in xs1]
plt.xticks(locs, keys)
ax = plt.gca()
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.set_ylim([0.65, 0.85])
plt.legend(loc='upper right')
plt.savefig(f"plots/bar_aurocs.pdf")
plt.close()
plt.bar(xs1, single_auprcs, color ='#FF1F5B', width = barWidth)
plt.bar(xs2, multi_auprcs, color ='#009ADE', width = barWidth)
plt.errorbar(xs1, single_auprcs, color ='black', yerr=single_auprcs_errors, fmt='o')
plt.errorbar(xs2, multi_auprcs, color ='black', yerr=multi_auprcs_errors, fmt='o')
plt.ylabel('AUPRC')#, fontweight ='bold', fontsize = 15)
locs = [(r + barWidth/2) for r in xs1]
plt.xticks(locs, keys)
ax = plt.gca()
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.set_ylim([0.35, 0.6])
plt.savefig(f"plots/bar_auprcs.pdf")
plt.close()
plot_types_bar()
def parse_results(task='pheno', file='results_val.txt'):
results_auroc = {}
results_auprc = {}
paths = glob.glob(f'checkpoints/{task}/ablation2/**/{file}', recursive = True)
for path in paths:
file1 = open(path, 'r')
Lines = file1.readlines()
bestlines = [line for line in range(len(Lines)) if 'best' in Lines[line] ]
lines_Selected = Lines[bestlines[-1]]
splited = [(num.split(':')[-1]) for num in lines_Selected.strip().split(' ')]
splited = [split for split in splited if split !='' and split[0]=='0']
print(splited)
# \, splited[5], splited[7])
auroc = float(splited[1])
auprc = float(splited[2])
ratio = float(path.split('/')[-2].split('_')[-1])
results_auroc[ratio] = auroc
results_auprc[ratio] = auprc
keys = results_auprc.keys()
keys = sorted(list(keys))
aurocs = [results_auroc[key] for key in keys]
auprcs = [results_auprc[key] for key in keys]
return aurocs, auprcs, keys
def plot_aurocs():
aurocs_mor = [0.868, 0.884, 0.872, 0.873, 0.874, 0.873, 0.876, 0.874, 0.875, 0.872, 0.872]
aurocs_mor_error = np.abs(np.array([(0.817, 0.901), (0.842, 0.919), (0.833, 0.910), (0.835, 0.906), (0.831, 0.910), (0.812, 0.911), (0.840, 0.910), (0.840, 0.911), (0.830, 0.911), (0.835, 0.906), (0.840, 0.911) ]).T - aurocs_mor)
# auprcs_mor_error = [(0.477, 0.714), (0.484, 0.718), (0.468, 0.714), (0.475, 0.703), (0.481, 0.708), (0.458, 0.695), (0.469, 0.701), (0.453, 0.681), (0.452, 0.689), (0.436, 0.669), (0.427, 0.660) ]
# auprcs_pheno = [0.492, 0.491, 0.496, 0.497, 0.495, 0.483, 0.489, 0.476, 0.466, 0.476, 0.466]
aurocs_pheno = [0.775, 0.773, 0.780, 0.776, 0.776, 0.775, 0.776, 0.771, 0.765, 0.771, 0.765]
aurocs_pheno_error = np.abs(np.array([(0.736, 0.812),(0.740, 0.810), (0.742, 0.820) , (0.739, 0.816), (0.730, 0.810), (0.736, 0.812), (0.737, 0.812), (0.732, 0.808), (0.726, 0.803), (0.732, 0.808), (0.726, 0.801)]).T - aurocs_pheno)
# plt.rcParams['figure.figsize'] = [8.0, 3.4]
plt.rcParams.update({'font.size': 12})
keys = np.array([0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7,0.8, 0.9, 1.0])*100
plt.plot(keys, aurocs_mor , label = f"In-hospital mortality", color='#FF1F5B')
plt.plot(keys, aurocs_pheno , label = f"Phenotyping", color='#009ADE')
plt.errorbar(keys, aurocs_mor, color ='#FF1F5B', yerr=aurocs_mor_error, fmt='*')
plt.errorbar(keys, aurocs_pheno, color ='#009ADE', yerr=aurocs_pheno_error, fmt='o')
ax = plt.gca()
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.set_ylim([0.70, 0.92])
plt.xticks(keys)
plt.yticks([0.70, 0.75, 0.80, 0.85, 0.90])
plt.ylabel('AUROC')
plt.xlabel('% of uni-modal training samples')
# plt.legend()
plt.legend(loc=(.6, .98))
plt.savefig(f"plots/aurocs.pdf")
plt.close()
plot_aurocs()
def plot_loss():
save_dir = 'checkpoints/mmtm_jointonly_a_e2c_on_p/'
checkpoint = torch.load(f'{save_dir}/last_checkpoint.pth.tar')
epochs_stats = checkpoint['epochs_stats']
colors = {
'loss train joint': '#8a2244',
'loss train': '#c687d5',
'loss train ehr': '#c687d5',
'loss train cxr': '#da8c22',
'loss val joint': '#80d6f8',
'loss val': '#440f06',
'loss val ehr' : '#440f06',
'loss val cxr': '#000075',
'auroc val ehr': '#02a92c',
'auroc val': '#02a92c',
'auroc val cxr': '#e6194B',
'auroc val joint': '#f58231',
'auroc val avg': '#ffe119',
# '#bfef45'
}
keys = [
'loss train joint',
'loss val joint',
'auroc val joint'
]
index = 1
values = ['loss', 'auroc']
filename = f'{values[index]}.png'
value = values[index]
for loss in epochs_stats:
if loss in keys and value in loss:
plt.plot(epochs_stats[loss], label = f"{loss.replace('avg', 'late')}", color=colors[loss])
plt.xlabel('epochs')
plt.title(value)
plt.legend()
plt.savefig(f"{save_dir}/{filename}")
plt.close()