-
Notifications
You must be signed in to change notification settings - Fork 42
/
Copy pathfeatures.py
219 lines (200 loc) · 7.33 KB
/
features.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
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
# -*- coding: utf-8 -*-
import pandas as pd
import time, datetime
import MySQLdb
from keras.utils import np_utils
import numpy as np
from sklearn import preprocessing
def date_difference(d1, d2):#d1早于d2
d1 = datetime.datetime.strptime(d1 + ' 00:00:00', '%Y-%m-%d %H:%M:%S')
d2 = datetime.datetime.strptime(d2 + ' 00:00:00', '%Y-%m-%d %H:%M:%S')
delta = d2 - d1
return delta.days
def extr_tags(x):
#type：国资系/上市公司系/银行系/民营系（类别型标签）
#tzzj_cooperation 投之家合作平台
#listed 股权上市
#vc 接受过风投
#argue 争议
#third_party 加入第三方征信
#join_society 加入协会
#label：停业/提现困难/跑路/经侦介入（含有这些关键字就为1，否则就为0）
types = ['国资系', '上市公司系', '银行系', '民营系']
binas = ['投之家合作平台', '股权上市', '接受过风投', '争议', '加入第三方征信', '加入协会']
labels = ['停业', '提现困难', '跑路','经侦介入']
rslt = [ [0 for i in xrange(0, len(x))] for j in xrange(0, 8)]
for i in xrange(0, len(x)):
tag_list = x[i].split(',')
for j in xrange(0, len(tag_list)):
tag = tag_list[j].encode('utf-8')
if tag in types:
rslt[0][i]=types.index(tag)+1
continue
if tag in binas:
rslt[binas.index(tag)+1][i] = 1
continue
if tag in labels:
rslt[7][i] = 1
return rslt
def one_feature(x, fill, cut):
x_new = []
for i in xrange(0, len(x)):
if x[i]=='':
x_new.append(0)
else:
x_new.append(x[i].strip(cut))
return x_new
def autobid_feature(x):
rslt = []
for i in xrange(0, len(x)):
if x[i]=='':
rslt.append(0)
elif x[i]==u'支持':
rslt.append(1)
else:
rslt.append(-1)
return rslt
def stockTransfer_feature(x):
rslt = []
for i in xrange(0, len(x)):
if x[i]=='':
rslt.append(-1)
elif x[i]==u'随时':
rslt.append(0)
elif x[i]==u'1年':
rslt.append(12)
elif x[i]==u'不可转让':
rslt.append(300)
else:
rslt.append(x[i].strip(u'个月'))
print x,
return rslt
def fundsToken_feature(x):
rslt = []
for i in xrange(0, len(x)):
if x[i]=='' or x[i]==u'无托管':
rslt.append(0)
else:
rslt.append(1)
return rslt
def ifGuarantee_feature(x):
rslt = []
for i in xrange(0, len(x)):
if x[i]=='':
rslt.append(0)
else:
rslt.append(1)
return rslt
def ifGuaranteeOrg_feature(x):
rslt = []
for i in xrange(0, len(x)):
if x[i]=='':
rslt.append(0)
else:
rslt.append(1)
return rslt
def lauchTime_feature(x):
rslt = []
for i in xrange(0, len(x)):
if x[i]!='':
rslt.append(date_difference(x[i], '2016-04-17'))
else:
rslt.append(0)
avg = np.mean(rslt)
for i in xrange(0, len(rslt)):
if rslt[i]==0:
rslt[i] = avg
rslt[i] = rslt[i]/30
return rslt
def category_feature(x):
dic = {'':-1, u'股份合作企业':0, u'私营企业':1, u'港、澳、台投资企业':2,
u'股份制企业':3, u'集体所有制企业':4, u'外商投资企业':5, u'国有企业':6, u'联营企业':7}
rslt = []
for i in xrange(0, len(x)):
rslt.append(dic[x[i]])
return rslt
conn = MySQLdb.connect(host="localhost",user="root",passwd="123",db="ppd",charset="utf8")
#平台基本信息
sql = "select id,platId,platName,platPin,tags,score,averageProfit,registMoney,\
autobid,stockTransfer,fundsToken,bidGuarantee,guaranteeMode,guaranteeOrg,\
lauchTime,category,lng,lat from platform"
platform = pd.read_sql(sql,conn,index_col="id")
#定义提取特征的DataFrame
platform_features = platform[['platId', 'platName', 'platPin']]
#开始提取特征
#tags特征
tags = list(platform['tags'])
tag_features = ['type', 'tzzj_cooperation', 'listed', 'vc', 'argue' ,'third_party', 'join_society', 'label']
rslt = extr_tags(tags)
for i in xrange(0, len(tag_features)):
platform_features[tag_features[i]] = pd.Series(rslt[i], index = platform_features.index)
#score
score = list(platform['score'])
platform_features['score'] = pd.Series(one_feature(score,0,''), index = platform_features.index)
#averageProfit
averageProfit = list(platform['averageProfit'])
platform_features['averageProfit'] = pd.Series(one_feature(averageProfit,0,'%'), index = platform_features.index)
#registMoney
registMoney = list(platform['registMoney'])
platform_features['registMoney'] = pd.Series(one_feature(registMoney,0,u' 万元'), index = platform_features.index)
#autobid
autobid = list(platform['autobid'])
platform_features['autobid'] = pd.Series(autobid_feature(autobid), index = platform_features.index)
#stockTransfer
stockTransfer = list(platform['stockTransfer'])
platform_features['stockTransfer'] = pd.Series(stockTransfer_feature(stockTransfer), index = platform_features.index)
#fundsToken
fundsToken = list(platform['fundsToken'])
platform_features['fundsToken'] = pd.Series(fundsToken_feature(fundsToken), index = platform_features.index)
#ifGuarantee
guaranteeMode = list(platform['guaranteeMode'])
platform_features['ifGuarantee'] = pd.Series(ifGuarantee_feature(guaranteeMode), index = platform_features.index)
#ifGuaranteeOrg
guaranteeOrg = list(platform['guaranteeOrg'])
platform_features['ifGuaranteeOrg'] = pd.Series(ifGuaranteeOrg_feature(guaranteeOrg), index = platform_features.index)
#lauchTime
lauchTime = list(platform['lauchTime'])
platform_features['lauchTime'] = pd.Series(lauchTime_feature(lauchTime), index = platform_features.index)
#category
category = list(platform['category'])
platform_features['category'] = pd.Series(category_feature(category), index = platform_features.index)
#lng
lng = list(platform['lng'])
platform_features['lng'] = pd.Series(one_feature(lng,0,''), index = platform_features.index)
#lat
lat = list(platform['lat'])
platform_features['lat'] = pd.Series(one_feature(lat,0,''), index = platform_features.index)
#用户评论信息comment
sql = "select timestamp,platName,attitude from comment"
comment = pd.read_sql(sql,conn)
platName = list(set(list(comment['platName'])))
comment_features = pd.DataFrame({'platName': pd.Series(platName)})
pos,neu,neg = [],[],[]
for i in xrange(0, len(platName)):
pos.append([0 for k in xrange(0, 12*2+3)])
neu.append([0 for k in xrange(0, 12*2+3)])
neg.append([0 for k in xrange(0, 12*2+3)])
tmp = comment[comment.platName==platName[i]]
tsp = list(tmp['timestamp'])
atti = list(tmp['attitude'])
for j in xrange(0, len(tsp)):
year = int(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(int(tsp[j]))).split(' ')[0].split('-')[0])
month = int(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(int(tsp[j]))).split(' ')[0].split('-')[1])
if year in [2014, 2015] or (year==2016 and month in [1, 2, 3]):
idx = (year-2014)*12+month-1
if atti[j]==u'推荐':
pos[i][idx]+=1
elif atti[j]==u'一般':
neu[i][idx]+=1
else:
neg[i][idx]+=1
pos[i] = ','.join([str(pos[i][j]) for j in xrange(0, len(pos[i]))])
neu[i] = ','.join([str(neu[i][j]) for j in xrange(0, len(neu[i]))])
neg[i] = ','.join([str(neg[i][j]) for j in xrange(0, len(neg[i]))])
comment_features['comment_pos'] = pd.Series(pos, index = comment_features.index)
comment_features['comment_neu'] = pd.Series(neu, index = comment_features.index)
comment_features['comment_neg'] = pd.Series(neg, index = comment_features.index)
platform_features = platform_features.merge(comment_features, how='left', on='platName')
platform_features.fillna(','.join(['0' for i in xrange(0, 12*2+3)]), inplace=True)
#把特征写入数据库
platform_features.to_sql('platform_features1', conn, flavor='mysql', if_exists='replace', index=True)