-
Notifications
You must be signed in to change notification settings - Fork 84
/
feat_a.py
306 lines (228 loc) · 15.4 KB
/
feat_a.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
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
from tool import *
cache_path = '/'
inplace = False
############################### 工具函数 ###########################
# 合并节约内存
def concat(L):
result = None
for l in L:
if result is None:
result = l
else:
result[l.columns.tolist()] = l
return result
# 统计转化率
def bys_rate(data,cate,cate2,label):
temp = data.groupby(cate2,as_index=False)[label].agg({'count':'count','sum':'sum'}).rename(columns={'2_total_fee':'1_total_fee'})
temp['rate'] = temp['sum']/temp['count']
data_temp = data[[cate]].copy()
data_temp = data_temp.merge(temp[[cate,'rate']],on=cate,how='left')
return data_temp['rate']
# 统计转化率
def mul_rate(data,cate,label):
temp1 = data.groupby([cate,label],as_index=False).size().unstack().fillna(0)
temp2 = data.groupby([cate], as_index=False).size()
temp2.loc[temp2 < 20] = np.nan
temp3 = (temp1.T/temp2).T
temp3.columns = [cate+'_'+str(c)+'_conversion' for c in temp3.columns]
temp3 = temp3.reset_index()
data = data.merge(temp3,on=cate,how='left')
return data
# 相同的个数
def get_same_count(li):
return pd.Series(li).value_counts().values[0]
# 相同的个数
def get_second_min(li):
return sorted(li)[1]
# One-hot encoding for categorical columns with get_dummies
def one_hot_encoder(df, nan_as_category=True, min_count=100,inplace=True):
original_columns = list(df.columns)
categorical_columns = [col for col in df.columns if df[col].dtype == 'object']
result = pd.get_dummies(df, columns=categorical_columns, dummy_na=nan_as_category)
new_columns = [c for c in result.columns if c not in original_columns]
cat_columns = [c for c in original_columns if c not in result.columns]
if not inplace:
for c in cat_columns:
result[c] = df[c]
for c in new_columns:
if (result[c].sum()<100) or ((result.shape[0]-result[c].sum())<100):
del result[c]
new_columns.remove(c)
return result, new_columns
# 连续特征离散化
def one_hot_encoder_continus(df, col, n_scatter=10,nan_as_category=True):
df[col+'_scatter'] = pd.qcut(df[col],n_scatter)
result = pd.get_dummies(df, columns=[col+'_scatter'], dummy_na=nan_as_category)
return result
############################### 预处理函数 ###########################
def pre_treatment(data,data_key):
result_path = cache_path + 'data_{}.feature'.format(data_key)
if os.path.exists(result_path) & 0:
data = pd.read_feature(result_path)
else:
month_fee = ['1_total_fee', '2_total_fee', '3_total_fee', '4_total_fee']
data['total_fee_mean4'] = data[month_fee[:4]].mean(axis=1)
data['total_fee_mean3'] = data[month_fee[:3]].mean(axis=1)
data['total_fee_mean2'] = data[month_fee[:2]].mean(axis=1)
data['total_fee_std4'] = data[month_fee[:4]].std(axis=1)
# data['total_fee_mode4'] = data[month_fee[:4]].apply(mode,axis=1)
data['total_fee_Standardization'] = data['total_fee_std4'] / (data['total_fee_mean4'] + 0.1)
data['1_total_fee_rate12'] = data['1_total_fee'] / (data['2_total_fee'] + 0.1)
data['1_total_fee_rate23'] = data['2_total_fee'] / (data['3_total_fee'] + 0.1)
data['1_total_fee_rate34'] = data['3_total_fee'] / (data['4_total_fee'] + 0.1)
data['1_total_fee_rate24'] = data['total_fee_mean2'] / (data['total_fee_mean4'] + 0.1)
data['total_fee_max4'] = data[month_fee[:4]].max(axis=1)
data['total_fee_min4'] = data[month_fee[:4]].min(axis=1)
data['total_fee_second_min4'] = data[month_fee[:4]].apply(get_second_min,axis=1)
data['service_caller_time_diff'] = data['service2_caller_time'] - data['service1_caller_time']
data['service_caller_time_sum'] = data['service2_caller_time'] + data['service1_caller_time']
data['service_caller_time_min'] = data[['service1_caller_time','service2_caller_time']].min(axis=1)
data['service_caller_time_max'] = data[['service1_caller_time', 'service2_caller_time']].max(axis=1)
data['1_total_fee_last0_number'] = count_encoding(data['1_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-1]).astype(int))
data['1_total_fee_last1_number'] = count_encoding(data['1_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-2]).astype(int))
data['1_total_fee_last2_number'] = count_encoding(data['1_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-4]).astype(int))
data['1_total_fee_last3_number'] = count_encoding(data['1_total_fee'].fillna(-1)//10)
data['2_total_fee_last0_number'] = count_encoding(data['2_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-1]).astype(int))
data['2_total_fee_last1_number'] = count_encoding(data['2_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-2]).astype(int))
data['2_total_fee_last2_number'] = count_encoding(data['2_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-4]).astype(int))
data['2_total_fee_last3_number'] = count_encoding(data['2_total_fee'].fillna(-1) // 10)
data['3_total_fee_last0_number'] = count_encoding(data['3_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-1]).astype(int))
data['3_total_fee_last1_number'] = count_encoding(data['3_total_fee'].fillna(-1).apply( lambda x:('%.2f' % x)[-2]).astype(int))
data['3_total_fee_last2_number'] = count_encoding(data['3_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-4]).astype(int))
data['3_total_fee_last3_number'] = count_encoding(data['3_total_fee'].fillna(-1) // 10)
data['4_total_fee_last0_number'] = count_encoding(data['4_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-1]).astype(int))
data['4_total_fee_last1_number'] = count_encoding(data['4_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-2]).astype(int))
data['4_total_fee_last2_number'] = count_encoding( data['4_total_fee'].fillna(-1).apply(lambda x: ('%.2f' % x)[-4]).astype(int))
data['4_total_fee_last3_number'] = count_encoding(data['4_total_fee'].fillna(-1) // 10)
# data['total_fee_sample_count'] = data[month_fee].apply(get_same_count,axis=1)
for fee in ['1_total_fee','2_total_fee', '3_total_fee', '4_total_fee']:
data['{}_1'.format(fee)] = ((data[fee] % 1==0) & (data[fee] !=0))
data['{}_01'.format(fee)] = ((data[fee] % 0.1==0) & (data[fee] !=0))
data['pay_number_last_2'] = data['pay_num']*100%100
# data = one_hot_encoder_continus(data,'1_total_fee',20)
# data = one_hot_encoder_continus(data, '2_total_fee', 20)
# data = one_hot_encoder_continus(data, '3_total_fee', 20)
# data = one_hot_encoder_continus(data, '4_total_fee', 20)
# data = one_hot_encoder_continus(data, 'age', 10)
# data = one_hot_encoder_continus(data, 'online_time', 10)
data['1_total_fee_log'] = np.log(data['1_total_fee']+2)
data['2_total_fee_log'] = np.log(data['2_total_fee'] + 2)
data['3_total_fee_log'] = np.log(data['3_total_fee'] + 2)
data['4_total_fee_log'] = np.log(data['4_total_fee'] + 2)
data = grp_standard(data, 'contract_type', ['1_total_fee_log'], drop=False)
data = grp_standard(data, 'contract_type', ['service_caller_time_min'], drop=False)
data = grp_standard(data, 'contract_type', ['service_caller_time_max'], drop=False)
data = grp_standard(data, 'contract_type', ['online_time'], drop=False)
data = grp_standard(data, 'contract_type', ['age'], drop=False)
data = grp_standard(data, 'net_service', ['1_total_fee_log'], drop=False)
data = grp_standard(data, 'net_service', ['service_caller_time_min'], drop=False)
data = grp_standard(data, 'net_service', ['service_caller_time_max'], drop=False)
data = grp_standard(data, 'net_service', ['online_time'], drop=False)
data = grp_standard(data, 'net_service', ['age'], drop=False)
data['age_scatter'] = pd.qcut(data['age'], 5)
data = grp_standard(data, 'age_scatter', ['1_total_fee_log'], drop=False)
data = grp_standard(data, 'age_scatter', ['service_caller_time_min'], drop=False)
data = grp_standard(data, 'age_scatter', ['service_caller_time_max'], drop=False)
data = grp_standard(data, 'age_scatter', ['online_time'], drop=False)
data = grp_standard(data, 'age_scatter', ['age'], drop=False)
data['online_time_scatter'] = pd.qcut(data['online_time'], 5)
data = grp_standard(data, 'online_time_scatter', ['1_total_fee_log'], drop=False)
data = grp_standard(data, 'online_time_scatter', ['service_caller_time_min'], drop=False)
data = grp_standard(data, 'online_time_scatter', ['service_caller_time_max'], drop=False)
data = grp_standard(data, 'online_time_scatter', ['online_time'], drop=False)
data = grp_standard(data, 'online_time_scatter', ['age'], drop=False)
data = grp_standard(data, 'service_type', ['1_total_fee_log'], drop=False)
data = grp_standard(data, 'service_type', ['service_caller_time_min'], drop=False)
data = grp_standard(data, 'service_type', ['service_caller_time_max'], drop=False)
data = grp_standard(data, 'service_type', ['online_time'], drop=False)
data = grp_standard(data, 'service_type', ['age'], drop=False)
del data['1_total_fee_log'],data['2_total_fee_log'],data['3_total_fee_log'],data['4_total_fee_log'], \
data['age_scatter'],data['online_time_scatter']
# data['online_time_count'] = count_encoding(data['online_time']//3)
data['month_traffic_last_month_traffic_sum'] = data['month_traffic'] + data['last_month_traffic']
data['month_traffic_last_month_traffic_diff'] = data['month_traffic'] - data['last_month_traffic']
data['month_traffic_last_month_traffic_rate'] = data['month_traffic'] / (data['last_month_traffic']+0.01)
data['outer_trafffic_month'] = data['month_traffic'] - data['local_trafffic_month']
data['local_trafffic_month_month_traffic_rate'] = data['local_trafffic_month'] / (data['month_traffic'] + 0.01)
data['month_traffic_last_month_traffic_sum_1_total_fee_rate'] = data['month_traffic_last_month_traffic_sum'] / (data['1_total_fee'] + 0.01)
data['month_traffic_local_caller_time'] = data['month_traffic'] / (data['local_caller_time'] + 0.01)
data['pay_num_per'] = data['pay_num'] / (data['pay_times']+0.01)
data['total_fee_mean4_pay_num_rate'] = data['pay_num'] / (data['total_fee_mean4'] + 0.01)
data['local_trafffic_month_spend'] = data['local_trafffic_month'] - data['last_month_traffic']
data['month_traffic_1_total_fee_rate'] = data['month_traffic'] / (data['1_total_fee'] + 0.01)
for traffic in ['month_traffic','last_month_traffic', 'local_trafffic_month']:
data['{}_1'.format(traffic)] = ((data[traffic] % 1==0) & (data[traffic] !=0))
data['{}_50'.format(traffic)] = ((data[traffic] % 50==0) & (data[traffic] !=0))
data['{}_1024'.format(traffic)] = ((data[traffic] % 1024==0) & (data[traffic] !=0))
data['{}_1024_50'.format(traffic)] = ((data[traffic] % 1024 % 50 == 0) & (data[traffic] != 0))
data['service_caller_time'] = data['service1_caller_time'] + data['service2_caller_time']
data['outer_caller_time'] = data['service_caller_time'] - data['local_caller_time']
data['local_caller_time_rate'] = data['local_caller_time'] / (data['service_caller_time']+0.01)
data['service1_caller_time_rate'] = data['service1_caller_time'] / (data['service_caller_time'] + 0.01)
data['local_caller_time_service2_caller_time_rate'] = data['local_caller_time'] / (data['service2_caller_time'] + 0.01)
data['service1_caller_time_1_total_fee_rate'] = data['service_caller_time'] / (data['1_total_fee'] + 0.01)
# data['online_fee'] = groupby(data,data,'online_time','total_fee_mean4','median')
# data['1_total_fee_10'] = data['1_total_fee']//10
# data['1_total_fee_10_online_time'] = groupby(data, data, '1_total_fee_10', 'online_time', 'median')
# del data['1_total_fee_10']
# data['per_month_fee'] = data['pay_num'] / (data['online_time']+0.01)
# data['per_month_times'] = data['pay_times'] / (data['online_time'] + 0.01)
# data
data['contract_time_count'] = count_encoding(data['contract_time'])
data['pay_num_count'] = count_encoding(data['pay_num'])
data['pay_num_last0_number'] = count_encoding(data['pay_num'].apply(lambda x: ('%.2f' % x)[-1]).astype(int))
data['pay_num_last1_number'] = count_encoding(data['pay_num'].apply(lambda x: ('%.2f' % x)[-2]).astype(int))
data['pay_num_last2_number'] = count_encoding(data['pay_num'].apply(lambda x: ('%.2f' % x)[-4]).astype(int))
data['pay_num_count'] = count_encoding(data['pay_num'] // 10)
data['age_count3'] = count_encoding(data['age'] // 3)
data['age_count6'] = count_encoding(data['age'] // 6)
data['age_count10'] = count_encoding(data['age'] // 10)
# data['contract_time_count'] = count_encoding(data['contract_time'])
# for i in range(11):
# data['temp'] = (data['label']==i).astype(int)
# data['1_total_fee_rate_cate{}'.format(i)] = cv_convert(data['1_total_fee'],data['temp'])
# del data['temp']
# data['1_total_fee_zheng'] = round(data['1_total_fee'])
# data = one_hot_encoder(data, '1_total_fee_zheng', n=4000, nan_as_category=True)
# 转化率
data = mul_rate(data, 'pay_num', 'current_service')
data = pd.get_dummies(data, columns=['contract_type'], dummy_na=-1)
data = pd.get_dummies(data, columns=['net_service'], dummy_na=-1)
data = pd.get_dummies(data, columns=['complaint_level'], dummy_na=-1)
data.reset_index(drop=True,inplace=True)
# data.to_feather(result_path)
return data
############################### 特征函数 ###########################
# 特征
def get__feat(data,data_key):
result_path = cache_path + '_feat_{}.feature'.format(data_key)
if os.path.exists(result_path) & (not inplace):
feat = pd.read_feature(result_path)
else:
data_temp = data.copy()
feat.to_feather(result_path)
return feat
# 二次处理特征
def second_feat(result):
return result
def make_feat(data,data_key):
t0 = time.time()
# data_key = hashlib.md5(data.to_string().encode()).hexdigest()
# #print('数据key为：{}'.format(data_key))
result_path = cache_path + 'feat_set_{}.feature'.format(data_key)
if os.path.exists(result_path) & 0:
result = pd.read_feature(result_path, 'w')
else:
data = pre_treatment(data,'data_key')
result = [data]
# #print('开始构造特征...')
# result.append(get_context_feat()) # context特征
# result.append(get_user_feat()) # 用户特征
# result.append(get_item_feat()) # 商品特征
# result.append(get_shop_feat()) # 商店特征
#print('开始合并特征...')
result = concat(result)
result = second_feat(result)
#print('特征矩阵大小：{}'.format(result.shape))
#print('生成特征一共用时{}秒'.format(time.time() - t0))
return result