-
Notifications
You must be signed in to change notification settings - Fork 38
/
Copy pathlgb_cv.py
115 lines (107 loc) · 4.3 KB
/
lgb_cv.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
import numpy as np
import lightgbm as lgb
from sklearn.model_selection import StratifiedKFold
# LightGBM模型+交叉验证
params = {'boosting_type': 'gbdt',
'objective': 'binary',
'metric': 'binary_logloss',
'learning_rate': 0.001,
'num_leaves': 82,
'max_depth': 8,
'min_data_in_leaf': 64,
'min_child_weight':1.435,
'bagging_fraction': 0.785,
'feature_fraction': 0.373,
'bagging_freq': 22,
'reg_lambda': 0.065,
'reg_alpha': 0.797,
'min_split_gain': 0.350,
'nthread': 8,
'seed': 42,
'scale_pos_weight':1.15,
'verbose': -1}
def get_lgb_oof(params, x_train, y_train, x_test, n_splits):
n_train, n_test = x_train.shape[0], x_test.shape[0]
kf = StratifiedKFold(n_splits=n_splits, random_state=0)
oof_train = np.empty((n_train, ))
oof_test = np.empty((n_test, ))
oof_test_skf = np.empty((n_splits, n_test))
for i, (train_index, test_index) in enumerate(kf.split(x_train, y_train)):
kf_x_train = x_train[train_index]
kf_y_train = y_train[train_index]
kf_x_test = x_train[test_index]
kf_y_test = y_train[test_index]
train_matrix = lgb.Dataset(kf_x_train, label=kf_y_train)
valid_matrix = lgb.Dataset(kf_x_test, label=kf_y_test)
model = lgb.train(params,
train_set=train_matrix,
num_boost_round=80000,
valid_sets=valid_matrix,
verbose_eval=-1,
early_stopping_rounds=600)
oof_test_skf[i, :] = model.predict(x_test)
oof_test[:] = oof_test_skf.mean(axis=0)
return oof_test.reshape(-1, 1)
# 技巧提升
# 用一个较大的learning rate学习得到初始版本模型1；
# 用一个较小的learning rate在模型1上继续训练得到模型2；
params1 = {'boosting_type': 'gbdt',
'objective': 'binary',
'metric': 'binary_logloss',
'learning_rate': 0.025,
"feature_fraction":0.5,
"num_leaves": 200,
"lambda_l1":2,
"lambda_l2":2,
"learning_rate":0.01,
'min_child_samples': 50,
"bagging_fraction":0.7,
"bagging_freq":1,
'verbose': -1}
params2 = {'boosting_type': 'gbdt',
'objective': 'binary',
'metric': 'binary_logloss',
'learning_rate': 0.001,
'num_leaves': 82,
'max_depth': 8,
'min_data_in_leaf': 64,
'min_child_weight': 1.435,
'bagging_fraction': 0.785,
'feature_fraction': 0.373,
'bagging_freq': 22,
'reg_lambda': 0.065,
'reg_alpha': 0.797,
'min_split_gain': 0.350,
'nthread': 8,
'seed': 42,
'scale_pos_weight':1.15,
'verbose': -1}
def get_lgb_oof(params1, params2, x_train, y_train, x_test, n_splits):
n_train, n_test = x_train.shape[0], x_test.shape[0]
kf = StratifiedKFold(n_splits=n_splits)
oof_train = np.empty((n_train, ))
oof_test = np.empty((n_test, ))
oof_test_skf = np.empty((n_splits, n_test))
for i, (train_index, test_index) in enumerate(kf.split(x_train, y_train)):
kf_x_train = x_train[train_index]
kf_y_train = y_train[train_index]
kf_x_test = x_train[test_index]
kf_y_test = y_train[test_index]
train_matrix = lgb.Dataset(kf_x_train, label=kf_y_train)
valid_matrix = lgb.Dataset(kf_x_test, label=kf_y_test)
model1 = lgb.train(params1,
train_set=train_matrix,
num_boost_round=20000,
valid_sets=valid_matrix,
verbose_eval=-1,
early_stopping_rounds=200)
model2 = lgb.train(params2,
train_set=train_matrix,
num_boost_round=20000,
valid_sets=valid_matrix,
init_model=model1,
verbose_eval=-1,
early_stopping_rounds=200)
oof_test_skf[i, :] = model2.predict(x_test)
oof_test[:] = oof_test_skf.mean(axis=0)
return oof_test.reshape(-1, 1)