-
Notifications
You must be signed in to change notification settings - Fork 1
/
preprocess.py
204 lines (125 loc) · 5.27 KB
/
preprocess.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
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
# coding: utf-8
import pandas as pd
import numpy as np
pd.set_option('display.max_columns',100)
train1=pd.read_table('../data/round1_iflyad_train.txt')
test1=pd.read_table('../data/round1_iflyad_test_feature.txt')
train2=pd.read_table('../data/round2_iflyad_train.txt')
test2=pd.read_table('../data/round2_iflyad_test_feature.txt')
train=train1.append(train2).drop_duplicates().reset_index(drop=True)
test=test1.append(test2).drop_duplicates().reset_index(drop=True)
del train['os']
del test['os']
train=train.sort_values(by='time').reset_index(drop=True)
test=test.sort_values(by='time').reset_index(drop=True)
test['click']=-1
concat=train.append(test).reset_index(drop=True)
concat=concat.sort_values(by='time').reset_index(drop=True)
concat['advert_industry_inner0']=concat['advert_industry_inner'].apply(lambda x:x.split('_')[0])
concat['advert_industry_inner1']=concat['advert_industry_inner'].apply(lambda x:x.split('_')[1])
del concat['advert_industry_inner']
import time, datetime
concat['mon']=concat['time'].apply(lambda x:(time.localtime(x)).tm_mon)
concat['day']=concat['time'].apply(lambda x:(time.localtime(x)).tm_mday)
concat['hour']=concat['time'].apply(lambda x:(time.localtime(x)).tm_hour)
concat['wday']=concat['time'].apply(lambda x:(time.localtime(x)).tm_wday)
concat['timee']=concat['time']
del concat['time']
rsav=concat[['instance_id','creative_width','creative_height']]
rsav.columns=['instance_id','creative_width_org','creative_height_org']
rsav.to_hdf('../data/size.hdf',key='xunfei')
concat['user_tags']=concat['user_tags'].astype(str)
concat['user_tags_len']=concat['user_tags'].apply(lambda x:len(x.split(',')))
alltags=concat['user_tags'].apply(lambda x:((x.split(','))))
alltagss=alltags.apply(lambda x:' '.join(x))
del concat['app_paid']
del concat['creative_is_voicead']
del concat['creative_is_js']
cool=list(concat.select_dtypes(include='bool').columns)
concat[cool]=(concat[cool]*1).values
#处理手机厂家，型号
concat['make']=concat['make'].apply(lambda x:str(x).lower())
concat['model']=concat['model'].apply(lambda x:str(x).lower())
def complete_col(xh,source='model',tarcol='make'):
#print(xh)
x=concat.loc[xh]
if (x[tarcol]=='nan'):
if x[source]!='nan':
if len(x[source].split())>0:
return x[source].split()[0]
else:
return 'nan'
else:
return 'nan'
else :
return x[tarcol]
makecom=pd.Series(range(len(concat))).apply(lambda x:complete_col(x,source='model',tarcol='make'))
concat['make']=makecom
concat['make']=concat['make'].apply(lambda x:x.replace('%2522',' ').strip())
def get_make(x):
#print(x)
if '%' in x:
try :return x.split('%')[0].split()[0]
except:return 'nan'
elif ',' in x:
try :return x.split(',')[0].split()[0]
except:return 'nan'
elif '+' in x:
try:return x.split('+')[0].split()[0]
except:return 'nan'
elif '-' in x:
try:return x.split('-')[0].split()[0]
except:return 'nan'
elif '.' in x:
try:return x.split('.')[0].split()[0]
except:return 'nan'
else :
try:return x.split()[0]
except:return 'nan'
concat['make']=concat['make'].apply(lambda x:get_make(x))
concat['make']=concat['make'].replace('mi','xiaomi').replace('sm','samsung')
def get_mke(x):
for mak in make_dict:
if mak in x:
return mak
return x
make_dict=list(concat['make'].value_counts()[:100].index)
concat['make']=concat['make'].apply(lambda x:get_mke(x))
tmp=concat['make'].value_counts()==1
other=list(tmp[tmp>0].index)
concat['make']=concat['make'].apply(lambda x:x if x not in other else 'other')
concat['model']=concat['model'].apply(lambda x:x.strip().replace('+',' ').replace(',','').replace('-',' ').replace('.',''))
tmp=concat['model'].value_counts()==1
other=list(tmp[tmp>0].index)
concat['model']=concat['model'].apply(lambda x:x if x not in other else 'other')
# 处理osv
import re
concat['osv']=concat['osv'].fillna('nan9998').astype(str)
concat['osv']=concat['osv'].apply(lambda x:re.sub(r'\D', "", (x)))
tmp=concat['osv'].value_counts()==1
other=list(tmp[tmp>0].index)
concat['osv']=concat['osv'].apply(lambda x:x if x not in other else 'other')
catcol=list(concat.loc[:,:'advert_name'])+['advert_industry_inner0','advert_industry_inner1']
catcol.remove('user_tags')
catcol.remove('instance_id')
from sklearn.preprocessing import LabelEncoder
enc = LabelEncoder()
for col in catcol:
concat[col]=concat[col].astype(str)
concat[col]=enc.fit_transform(concat[col].fillna(-1))
cool=list(concat.select_dtypes(include='int64').columns)
cool.remove('instance_id')
cool.remove('timee')
concat[cool]=concat[cool].astype(np.int32)
del concat['advert_name']
concat['user_tagss']=concat['user_tags']
del concat['user_tags']
#处理user_tags
concat['user_tagss']=concat['user_tagss'].apply(lambda x:x.split(',')).apply(lambda x:' '.join(x)).apply(lambda x:x.split()).apply(lambda x:' '.join(x))
concat['target']=concat['click']
del concat['click']
train_data = concat[concat['target']!=-1]
test_data = concat[concat['target']==-1]
train_data.to_hdf('../data/train_data2.hdf', key='xunfei')
test_data.to_hdf('../data/test_data2.hdf', key='xunfei')
concat['instance_id'].to_hdf('../data/allid2.hdf', key='xunfei')