-
Notifications
You must be signed in to change notification settings - Fork 1
/
gen_result.py
187 lines (154 loc) · 7.35 KB
/
gen_result.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
import pandas as pd
import sys, pickle, string, os
import xgboost as xgb
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import ExtraTreesClassifier
from sklearn.ensemble import RandomForestRegressor
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.linear_model import LogisticRegression, LinearRegression, LogisticRegressionCV
from sklearn.preprocessing import StandardScaler
from subprocess import call
from sklearn.datasets import dump_svmlight_file
from sklearn.cross_validation import ShuffleSplit
predictions = {}
for r in range(40):
print r
train_data = pd.read_csv("data/train.csv")
test_data = pd.read_csv("data/test.csv")
test_user=test_data['user']
test_merchant=test_data['merchant']
split = ShuffleSplit(train_data.shape[0], n_iter = 1, test_size=0.16)
for tr, te in split:
train1, train2 = tr, te
train1_data = train_data.iloc[train1,:]
train2_data = train_data.iloc[train2,:]
train_label1 = train1_data['label']
train_label2 = train2_data['label']
train_label = train_data['label']
#print train_label
del train_data['label']
del train1_data['label']
del train2_data['label']
del test_data['label']
del test_data['user']
del test_data['merchant']
dtrain1 = xgb.DMatrix( train1_data.values, label=train_label1.values)
dtrain2 = xgb.DMatrix( train2_data.values, label=train_label2.values)
dtest = xgb.DMatrix( test_data.values)
print "training gbc:tree boosters"
param = {'bst:max_depth':3, 'bst:eta':0.1, 'silent':1, 'objective':'binary:logistic', 'nthread' : 8, 'eval_metric':'auc' }
num_round = 125
bst = xgb.train( param, dtrain1, num_round)
pred_label_train = bst.predict(dtrain2)
pred_label_test = bst.predict( dtest )
print "training extra trees forest"
clf = ExtraTreesClassifier(n_estimators=1000, n_jobs=8,min_samples_split=10)
clf.fit(train1_data, train_label1)
pred_label_train2 = clf.predict_proba(train2_data)[:,1]
pred_label_test2 = clf.predict_proba(test_data)[:,1]
pred_train = np.hstack(( np.reshape(pred_label_train,(-1,1)), np.reshape(pred_label_train2,(-1,1)) ))
pred_test = np.hstack(( np.reshape(pred_label_test,(-1,1)), np.reshape(pred_label_test2,(-1,1)) ))
'''
print "random forest trees forest"
clf = RandomForestClassifier(n_estimators=100, n_jobs=8,min_samples_split=6)
clf.fit(train1_data, train_label1)
pred_label_train4 = clf.predict_proba(train2_data)[:,1]
pred_label_test4 = clf.predict_proba(test_data)[:,1]
pred_train = np.hstack(( pred_train, np.reshape(pred_label_train4,(-1,1)) ))
pred_test = np.hstack(( pred_test, np.reshape(pred_label_test4,(-1,1)) ))
'''
print "Gradient Boosting Classifier"
clf = GradientBoostingClassifier(n_estimators=1000, min_samples_split=10)
clf.fit(train1_data, train_label1)
pred_label_train6 = clf.predict_proba(train2_data)[:,1]
pred_label_test6 = clf.predict_proba(test_data)[:,1]
pred_train = np.hstack(( np.reshape(pred_label_train,(-1,1)), np.reshape(pred_label_train6,(-1,1)) ))
pred_test = np.hstack(( np.reshape(pred_label_test,(-1,1)), np.reshape(pred_label_test6,(-1,1)) ))
#pred_train = np.hstack(( pred_train, np.reshape(pred_label_train6,(-1,1)) ))
#pred_test = np.hstack(( pred_test, np.reshape(pred_label_test6,(-1,1)) ))
#vw
print "Vowpal Wabbit"
ss = StandardScaler()
train1_data_norm = ss.fit_transform(train1_data)
train2_data_norm = ss.transform(train2_data)
test_data_norm = ss.transform(test_data)
ntrain1_label = train_label1.copy()
ntrain1_label.values[np.where(ntrain1_label == 0)] = -1
ntrain2_label = train_label2.copy()
ntrain2_label.values[np.where(ntrain2_label == 0)] = -1
dump_svmlight_file(train1_data, ntrain1_label, "train1.vw", zero_based=False)
dump_svmlight_file(train2_data, ntrain2_label,"train2.vw", zero_based=False)
dump_svmlight_file(test_data, np.zeros((test_data_norm.shape[0],)), "test.vw", zero_based=False)
print 1
of = open( "vw_train1set.csv" ,"w")
of2 = open("vw_train2set.csv" ,"w")
of3 = open("vw_testset.csv" ,"w")
fi = open("train1.vw","r")
for lines in fi:
li = lines.strip().split()
of.write( li[0] )
of.write(" | ")
of.write( string.join(li[1:]," "))
of.write("\n")
of.close()
fi = open("train2.vw", "r")
for lines in fi:
li = lines.strip().split()
of2.write( li[0] )
of2.write(" | ")
of2.write( string.join(li[1:]," "))
of2.write("\n")
of2.close()
fi = open("test.vw","r")
for lines in fi:
li = lines.strip().split()
of3.write( "1")
of3.write(" | ")
of3.write( string.join(li[1:]," "))
of3.write("\n")
of3.close()
call("vw -c /Users/shen/Desktop/newtry/vw_train1set.csv --passes 100 --loss_function=logistic -f /Users/shen/Desktop/newtry/vw_trainset_model.vw",shell=True)
call("vw /Users/shen/Desktop/newtry/vw_train2set.csv -t -i /Users/shen/Desktop/newtry/vw_trainset_model.vw --loss_function=logistic --link=logistic -p /Users/shen/Desktop/newtry/vwpreds_train2.txt", shell=True)
call("vw /Users/shen/Desktop/newtry/vw_testset.csv -t -i /Users/shen/Desktop/newtry/vw_trainset_model.vw --loss_function=logistic --link=logistic -p /Users/shen/Desktop/newtry/vwpreds_test.txt", shell=True)
pred_label_train7 = []
fi = open( "vwpreds_train2.txt" ,"r")
for lines in fi:
li = lines.strip().split()
pred_label_train7.append(float(li[0]))
pred_label_test7 = []
fi.close()
fi = open( "vwpreds_test.txt" ,"r")
for lines in fi:
li = lines.strip().split()
pred_label_test7.append(float(li[0]))
pred_train = np.hstack(( pred_train, np.reshape(pred_label_train7,(-1,1)) ))
pred_test = np.hstack(( pred_test, np.reshape(pred_label_test7,(-1,1)) ))
fi.close()
#blend
#print pred_train234
clf = GradientBoostingClassifier(n_estimators=25, min_samples_split=10)
clf.fit(pred_train, train_label2)
pred_label_test = clf.predict_proba(pred_test)[:,1]
'''
dtrain2 = xgb.DMatrix(pred_train, label=train_label2.values)
dtest = xgb.DMatrix( pred_test )
print "training blend : xgb trees booster logistic regression, max depth 2"
param = {'bst:max_depth':2, 'bst:eta':0.1, 'silent':1, 'objective':'binary:logistic', 'nthread' : 8, 'eval_metric':'auc' }
num_round = 50
bst = xgb.train( param, dtrain2, num_round)
pred_label1 = bst.predict( dtest )
print "training blend : xgb linear booster logistic regression"
param = {'booster_type':1, 'bst:lambda':0, 'bst:alpha':0, 'bst:lambda_bias':0, 'silent':1, 'objective':'binary:logistic', 'nthread' : 8, 'eval_metric':'auc' }
num_round = 25
bst = xgb.train( param, dtrain2, num_round)
pred_label2 = bst.predict( dtest )
'''
predictions[r]=pred_label_test
#mean_pred = (pred_label1 + pred_label2)/2.
mean_pred = (predictions[0]+predictions[1]+predictions[2]+predictions[3]+predictions[4]+predictions[5]+predictions[6]+predictions[7]+predictions[8]+predictions[9]+predictions[10]+predictions[11]+predictions[12]+predictions[13]+predictions[14]+predictions[15]+predictions[16]+predictions[17]+predictions[18]+predictions[19]+predictions[20]+predictions[21]+predictions[22]+predictions[23]+predictions[24]+predictions[25]+predictions[26]+predictions[27]+predictions[28]+predictions[29]+predictions[30]+predictions[31]+predictions[32]+predictions[33]+predictions[34]+predictions[35]+predictions[36]+predictions[37]+predictions[38]+predictions[39])/40.
#mean_pred = (predictions[0]+predictions[1]+predictions[2])/3.
of = open("submission.csv","w")
of.write("user_id#merchant_id,prob\n")
for i in range(len(pred_label_test)):
of.write( str(test_user[i])+'#'+str(test_merchant[i])+','+str(mean_pred[i])+'\n' )