-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain_tm_model.py
134 lines (121 loc) · 4.73 KB
/
train_tm_model.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
import os
import pandas as pd
import numpy as np
import pickle
import sys
sys.path.append(os.path.expanduser("~/TMPredictor/survival_tm/auton-survival"))
from sklearn.model_selection import ParameterGrid
from auton_survival.estimators import SurvivalModel
import sys
dumping_dir = sys.argv[1]
x_dir = sys.argv[2]
e_dir = sys.argv[3]
t_dir = sys.argv[4]
print(f'x_dir: {x_dir}')
print(f'e_dir: {e_dir}')
print(f't_dir: {t_dir}')
with open(f'data/{x_dir}.pkl', 'rb') as f:
x_train_df = pd.DataFrame(pickle.load(f))
with open(f'data/{e_dir}.pkl', 'rb') as f:
e = np.array(pickle.load(f))
with open(f'data/{t_dir}.pkl', 'rb') as f:
t = np.array(pickle.load(f))
'''with open('data/x_train_memory_normed.pkl', 'rb') as f:
x_train_df = pickle.load(f)
with open('data/x_valid_memory_normed.pkl', 'rb') as f:
x_valid_df = pickle.load(f)
with open('data/outcomes_train_memory.pkl', 'rb') as f:
outcomes_train_df = pickle.load(f)
with open('data/outcomes_valid_memory.pkl', 'rb') as f:
outcomes_valid_df = pickle.load(f)'''
'''with open('data/x_train_df_normed.pkl', 'rb') as f:
x_train_df = pickle.load(f)
with open('data/x_valid_df_normed.pkl', 'rb') as f:
x_valid_df = pickle.load(f)
with open('data/outcomes_train_df.pkl', 'rb') as f:
outcomes_train_df = pickle.load(f)
with open('data/outcomes_valid_df.pkl', 'rb') as f:
outcomes_valid_df = pickle.load(f)'''
'''with open('data/x_train_future_normed.pkl', 'rb') as f:
x_train_df = pickle.load(f)
with open('data/x_valid_future_normed.pkl', 'rb') as f:
x_valid_df = pickle.load(f)
with open('data/outcomes_train_future.pkl', 'rb') as f:
outcomes_train_df = pickle.load(f)
with open('data/outcomes_valid_future.pkl', 'rb') as f:
outcomes_valid_df = pickle.load(f)'''
'''with open('data/x_train_pca.pkl', 'rb') as f:
x_train_df = pickle.load(f)
with open('data/x_valid_pca.pkl', 'rb') as f:
x_valid_df = pickle.load(f)
with open('data/outcomes_train_df.pkl', 'rb') as f:
outcomes_train_df = pickle.load(f)
with open('data/outcomes_valid_df.pkl', 'rb') as f:
outcomes_valid_df = pickle.load(f)'''
'''with open('data/x_train_pca.pkl', 'rb') as f:
x_train_df = pickle.load(f)
with open('data/x_valid_pca.pkl', 'rb') as f:
x_valid_df = pickle.load(f)
with open('data/outcomes_train_manual_norm.pkl', 'rb') as f:
outcomes_train_df = pickle.load(f)
with open('data/outcomes_valid_manual_norm.pkl', 'rb') as f:
outcomes_valid_df = pickle.load(f)'''
'''with open('data/rt_x_pca_bms.pkl', 'rb') as f:
x_train_df = pd.DataFrame(pickle.load(f))
with open('data/rt_filtered_e_bms.pkl', 'rb') as f:
e = pickle.load(f)
with open('data/rt_filtered_t_bms.pkl', 'rb') as f:
t = pickle.load(f)'''
'''with open('data/cakenn_x_pca_bms_2.pkl', 'rb') as f:
x_train_df = pd.DataFrame(pickle.load(f))
with open('data/cakenn_e_bms.pkl', 'rb') as f:
e = pickle.load(f)
with open('data/cakenn_t_bms.pkl', 'rb') as f:
t = pickle.load(f)'''
'''with open('data/cakenn_x_pca.pkl', 'rb') as f:
x_train_df = pd.DataFrame(pickle.load(f))
with open('data/cakenn_e.pkl', 'rb') as f:
e = pickle.load(f)
with open('data/cakenn_t.pkl', 'rb') as f:
t = pickle.load(f)'''
outcomes_train_df = pd.DataFrame({'time': t, 'event': e})
param_grid = {'k' : [3],
'iters': [2500],
'distribution' : ['LogNormal'],
'learning_rate' : [1e-5],
'batch_size' : [1000],
'layers' : [
[100, 1000]
]
}
'''param_grid = {'k' : [3, 6, 10],
'iters': [2500],
'distribution' : ['LogNormal'],
'learning_rate' : [1e-5],
'batch_size' : [1000],
'layers' : [
[100, 300, 1000],
[300, 300, 300],
[100, 1000],
[50, 80, 100, 150]
]
}'''
params = ParameterGrid(param_grid)
models=[]
for i, param in enumerate(params):
print('Hyperparameter ' + str(i) + ' of ' + str(len(params)))
print(param)
model = SurvivalModel(model='dsm',
iters=param['iters'],
k=param['k'],
layers=param['layers'],
distribution=param['distribution'],
learning_rate=param['learning_rate'],
batch_size=param['batch_size']
)
#_, train_loss, val_loss = model.fit(x_train_df, outcomes_train_df, val_data=(x_valid_df, outcomes_valid_df))
_, train_loss, val_loss = model.fit(x_train_df, outcomes_train_df)
models.append([model, train_loss, val_loss, param])
print(f'dumping to models/{dumping_dir}.pkl')
with open(f'models/{dumping_dir}.pkl', 'wb') as f:
pickle.dump(models, f)