Chem_Expl_Comp/train.py at main · TrinitroCat/Chem_Expl_Comp · GitHub

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
"""

"""
import time
import copy

import numpy as np
from sklearn.ensemble import RandomForestRegressor, GradientBoostingRegressor
from sklearn.neural_network import MLPRegressor
from sklearn.metrics import r2_score, mean_absolute_error, root_mean_squared_error
import joblib as jb
import torch as th

from ppx_tools.main import Select, Trans

from MLP_model import MLP, TrainMLP

from load_files import load_files

RFx = RandomForestRegressor(
    n_estimators=200,
    max_depth=18,
    min_samples_leaf=2,
    min_samples_split=2,
    verbose=0,
    n_jobs=-1
)
RFy = RandomForestRegressor(
    n_estimators=400,
    max_depth=16,
    min_samples_leaf=1,
    min_samples_split=2,
    verbose=0,
    n_jobs=-1
)

GBRx = GradientBoostingRegressor(
    learning_rate=0.1,
    n_estimators=300,
    verbose=1,
)
GBRy = copy.deepcopy(GBRx)

'''
MLPx = MLPRegressor(
    hidden_layer_sizes=(150,150,150,150),
    solver='adam',
    learning_rate='adaptive',
    learning_rate_init=0.0005,
    batch_size=275,
    max_iter=2000,
    n_iter_no_change=50,
    verbose=True
)
MLPy = copy.deepcopy(MLPx)
'''
# torch implementation
MLPx = MLP(20, 4, 100)
MLPy = MLP(20, 4, 100)

# load files
data = load_files('origin_dataset.xlsx', label_col=-2)
# classified by components
component_set = {data.samp_name[0, 1:].tobytes(), }
split_indices = list()
for i, name in enumerate(data.samp_name):
    hashName = name[1:].tobytes()
    if hashName not in component_set:
        component_set.add(hashName)
        split_indices.append(i)
# split
data.blocked_split(split_indices, True, 0.1, False, 0.1)

# feature engineering
'''trans = Trans(data.train_feat[:, :-1], data.train_label, data.feat_name[:-1])
X_trans, X_trans_name = trans.elem_func(name=True)
data.train_feat = np.concatenate([X_trans, data.train_feat[:, -1:]], axis=1)
data.train_names = np.concatenate([X_trans_name, data.feat_name[-1:]], axis=0)

sel = Select(data.train_feat, data.train_label[:, 0], data.feat_name)
X_select = sel.SIS('distance', size=300)
sel = Select(X_select, data.train_label[:, 0], data.feat_name)
sel.RFE(RFx, 1, )'''

# train
RFx.fit(data.train_feat, data.train_label[:, 0])
RFy.fit(data.train_feat, data.train_label[:, 1])
GBRx.fit(data.train_feat, data.train_label[:, 0])
GBRy.fit(data.train_feat, data.train_label[:, 1])
'''
MLPx.fit(data.train_feat, data.train_label[:, 0])
MLPy.fit(data.train_feat, data.train_label[:, 1])
'''
# torch train
print('\nTorch Training ...')
trainer_x = TrainMLP(MLPx, model_name='T')
trainer_x.run(2000, th.from_numpy(data.train_feat), th.from_numpy(data.train_label[:, 0]), save_thres=4.)
trainer_y = TrainMLP(MLPy, model_name='y')
trainer_y.run(2000, th.from_numpy(data.train_feat), th.from_numpy(data.train_label[:, 1]), save_thres=0.1)
print('Done. \n')

# validation
print('*'*120)
MODEL_NAME = ('RF', 'GBR', 'MLP')
for algo in (RFx, GBRx, MLPx):
    algo.verbose = 0
print('----- Temperature -----')
print('R^2:')
r2_list = list()
for i, algo in enumerate((RFx, GBRx, MLPx)):
    r2 = r2_score(data.val_label[:, 0], algo.predict(data.val_feat))
    r2_list.append(r2)
    print(f'\t{MODEL_NAME[i]:<5}: {r2:<5.3f}')
print('MAE:')
for i, algo in enumerate((RFx, GBRx, MLPx)):
    print(f'\t{MODEL_NAME[i]:<5}: {mean_absolute_error(data.val_label[:, 0], algo.predict(data.val_feat)):<6.4f}')
print('RMSE:')
for i, algo in enumerate((RFx, GBRx, MLPx)):
    rmse = root_mean_squared_error(data.val_label[:, 0], algo.predict(data.val_feat))
    print(f'\t{MODEL_NAME[i]:<5}: {rmse:<6.4f}')
    # Save models
    jb.dump(algo, f'trained_models/{MODEL_NAME[i]}x_RMSE_{rmse:.4f}_r2_{r2_list[i]:.3f}.pkl')
print('----- Y -----')
print('R^2:')
r2_list = list()
for i, algo in enumerate((RFy, GBRy, MLPy)):
    r2 = r2_score(data.val_label[:, 1], algo.predict(data.val_feat))
    r2_list.append(r2)
    print(f'\t{MODEL_NAME[i]:<5}: {r2:<5.3f}')
print('MAE:')
for i, algo in enumerate((RFy, GBRy, MLPy)):
    print(f'\t{MODEL_NAME[i]:<5}: {mean_absolute_error(data.val_label[:, 1], algo.predict(data.val_feat)):<6.4f}')
print('RMSE:')
for i, algo in enumerate((RFy, GBRy, )):
    rmse = root_mean_squared_error(data.val_label[:, 1], algo.predict(data.val_feat))
    print(f'\t{MODEL_NAME[i]:<5}: {rmse:<6.4f}')
    # Save models
    jb.dump(algo, f'trained_models/{MODEL_NAME[i]}y_RMSE_{rmse:.4f}_r2_{r2_list[i]:.3f}.pkl')