-
Notifications
You must be signed in to change notification settings - Fork 9
/
Copy pathcase_earthquake.py
188 lines (167 loc) · 7.52 KB
/
case_earthquake.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
# https://www.kaggle.com/tocha4/lanl-master-s-approach
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import scipy as sc
import matplotlib.pyplot as plt
import seaborn as sns
import gc
import warnings
warnings.filterwarnings("ignore")
warnings.simplefilter(action='ignore', category=FutureWarning)
from tqdm import tqdm_notebook
import datetime
import time
import random
from joblib import Parallel, delayed
import lightgbm as lgb
from tensorflow import keras
from gplearn.genetic import SymbolicRegressor
#from catboost import Pool, CatBoostRegressor
from litemort import *
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import mean_absolute_error,mean_squared_error
from sklearn.model_selection import GridSearchCV, KFold, RandomizedSearchCV
from sklearn.feature_selection import RFECV, SelectFromModel
import os
import sys
import pickle
from sklearn.linear_model import LinearRegression, Ridge
from sklearn.tree import DecisionTreeRegressor
from sklearn.svm import NuSVR, SVR
from sklearn.kernel_ridge import KernelRidge
from sklearn.ensemble import AdaBoostRegressor
from sklearn.ensemble import RandomForestRegressor, ExtraTreesRegressor
today = datetime.date.today().strftime('%m%d')
isMORT = len(sys.argv)>1 and sys.argv[1] == "mort"
#isMORT = True
#some_rows=3000
some_rows=None
model_type='mort' if isMORT else 'lgb'
nVerbose = 500
pkl_path = 'G:/kaggle/Earthquake/data/anton_2_{}.pickle'.format(some_rows)
pkl_path = 'G:/kaggle/Earthquake/data/anton_cys0_{}.pickle'.format(some_rows)
eval_metric='l1'
min_error = mean_squared_error if eval_metric=='l1' else mean_absolute_error
params = {
'n_estimators':50000, #减少n_estimators 并不能控制overfit
'early_stopping_rounds': 200,
'num_leaves': 256, #128
#'max_bin': 64,
'min_data_in_leaf': 32, #79
'objective': 'tweedie', #'regression',
'max_depth': -1,
'learning_rate': 0.01,
#"boosting": "gbdt",
"bagging_freq": 5,
"bagging_fraction": 1,#0.8126672064208567, #0.8126672064208567,
"bagging_seed": 11,
"metric": 'mae',
"verbosity": nVerbose,
#'reg_alpha': 0.1302650970728192,
#'reg_lambda': 0.3603427518866501,
'colsample_bytree': 0.05
}
print("params=\n{}\n".format(params))
submission = pd.read_csv('G:/kaggle/Earthquake/input/sample_submission.csv')
def Load_MoreDatas(paths):
train_s=[]
y_s=[]
for path,nFile in paths:
for i in range(nFile):
path_X,path_y="{}/train_X_features_{}.csv".format(path,i+1),"{}/train_y_{}.csv".format(path,i+1)
X_ = pd.read_csv(path_X)
y_ = pd.read_csv(path_y, index_col=False, header=None)
train_s.append(X_)
y_s.append(y_)
print("X_[{}]@{}\ny_[{}]@{}".format(X_.shape,path_X,y_.shape,path_y))
if len(train_s)>0:
train_X = pd.concat(train_s, axis=0)
y = pd.concat(y_s, axis=0)
train_X = train_X.reset_index(drop=True)
y = y.reset_index(drop=True)
print("Load_MoreDatas X_[{}] y_[{}]".format(train_X.shape, y.shape))
return train_X,y
if os.path.isfile(pkl_path):
print("\n======load pickle file from {} ...".format(pkl_path))
with open(pkl_path, "rb") as fp: # Pickling
[train_X, test_X, train_y] = pickle.load(fp)
if some_rows is not None:
train_X = train_X[:some_rows]
test_X = test_X[:some_rows]
train_y = train_y[:some_rows]
print("\n======train_X={} test_X={} train_y={} \n".format(train_X.shape, test_X.shape, train_y.shape))
else:
#train_X_2,y_2 = Load_MoreDatas([('G:/kaggle/Earthquake/data/cys/15000', 14),
# ('G:/kaggle/Earthquake/data/cys/17000', 15)])
train_X_0 = pd.read_csv("G:/kaggle/Earthquake/data/train_X_features_865_0.csv")
train_X_1 = pd.read_csv("G:/kaggle/Earthquake/data/train_X_features_865_1.csv")
y_0 = pd.read_csv("G:/kaggle/Earthquake/data/train_y_0.csv", index_col=False, header=None)
y_1 = pd.read_csv("G:/kaggle/Earthquake/data/train_y_1.csv", index_col=False, header=None)
train_X = pd.concat([train_X_0, train_X_1], axis=0)
y = pd.concat([y_0, y_1], axis=0)
train_X = train_X.reset_index(drop=True)
print(train_X.shape)
print(train_X.head())
y = y.reset_index(drop=True)
print(y[0].shape)
train_y = pd.Series(y[0].values)
test_X = pd.read_csv("G:/kaggle/Earthquake/data/test_X_features_10.csv")
scaler = StandardScaler()
train_columns = train_X.columns
train_X[train_columns] = scaler.fit_transform(train_X[train_columns])
test_X[train_columns] = scaler.transform(test_X[train_columns])
with open(pkl_path, "wb") as fp: # Pickling
pickle.dump([train_X, test_X, train_y], fp)
print("Save pickle file at {} train_X={} test_X={} train_y={}".format(pkl_path,train_X.shape, test_X.shape, train_y.shape))
sys.exit(-2)
train_columns = train_X.columns
n_fold = 5 #n_fold=10 只是增加了过拟合，莫名其妙
folds = KFold(n_splits=n_fold, shuffle=True, random_state=42)
oof = np.zeros(len(train_X))
train_score = []
fold_idxs = []
# if PREDICTION:
predictions = np.zeros(len(test_X))
feature_importance_df = pd.DataFrame()
#run model
for fold_, (trn_idx, val_idx) in enumerate(folds.split(train_X,train_y.values)):
t0=time.time()
strLog = "fold {}".format(fold_)
print(strLog)
fold_idxs.append(val_idx)
fold_importance_df = pd.DataFrame()
fold_importance_df["Feature"] = train_columns
X_train, X_valid = train_X[train_columns].iloc[trn_idx], train_X[train_columns].iloc[val_idx]
y_train, y_valid = train_y.iloc[trn_idx], train_y.iloc[val_idx]
if model_type == 'mort':
params['objective'] = 'regression'
# model = LiteMORT(params).fit(X_train, y_train, eval_set=[(X_valid, y_valid)])
model = LiteMORT(params).fit_1(X_train, y_train, eval_set=[(X_valid, y_valid)])
if model_type == 'cat':
model = CatBoostRegressor(n_estimators=25000, verbose=-1, objective="MAE", loss_function="MAE", boosting_type="Ordered", task_type="GPU")
model.fit(X_tr,
y_tr,
eval_set=[(X_val, y_val)],
# eval_metric='mae',
verbose=2500,
early_stopping_rounds=500)
if model_type == 'lgb':
model = lgb.LGBMRegressor(**params, n_jobs=-1)#n_estimators=50000,
model.fit(X_train, y_train,
eval_set=[(X_train, y_train), (X_valid, y_valid)], eval_metric='mae',
verbose=nVerbose, early_stopping_rounds=200) #
fold_importance_df["importance"] = model.feature_importances_[:len(train_columns)]
fold_importance_df["fold"] = fold_ + 1
feature_importance_df = pd.concat([feature_importance_df, fold_importance_df], axis=0)
oof[val_idx] = model.predict(X_valid)
fold_score = mean_absolute_error(oof[val_idx], y_valid)
print("{}\tscore={:.4g} time={:.4g}".format(strLog,fold_score,time.time()-t0))
#predictions
predictions += model.predict(test_X[train_columns]) / folds.n_splits
train_score.append(fold_score)
cv_score = mean_absolute_error(train_y, oof)
print(f"\n======After {n_fold} score = {cv_score:.3f}, CV_fold = {np.mean(train_score):.3f} | {np.std(train_score):.3f}", end=" ")
submission["time_to_failure"] = predictions
submission.to_csv(f'G:/kaggle/Earthquake/result/{model_type}_{today}_[{cv_score:.3f},{np.std(train_score):.3f}].csv', index=False)
submission.head()