新增设备信息特征6个

config/concat_sql.py

@@ -98,6 +98,11 @@ con_vcc_transfer = create_engine('mysql+pymysql://real_data_test:qpRBSarm$Tv*YO!
 
 con_vcc_transfer_pay = create_engine('mysql+pymysql://real_data_test:qpRBSarm$Tv*YO!n@172.30.220.16:33316/payment_center?charset=utf8',echo=False)
 
+con_device_info = create_engine('mysql+pymysql://pengcheng_zhang:phBnseTGbmUd7cKI@fengkong-tidb.quantgroups.com:4010/rc_user_behavior?charset=utf8'
+                                  ,echo=False)
+
+con_vcc_device_info = create_engine('mysql+pymysql://bigdata_test:ieneue7ldondk03d@172.30.5.27:8066/acsdb?charset=utf8'
+                                ,echo=False)
 # vcc_talos信息：
 # url:  vcc-db.quantgroups.com:3306
 # user： rc_slave_user_r

features/calc_features/DoubleLoan_v4.py

+# -*- coding:utf-8 -*-
+
+from features.Base_Features import BaseFeatures as BF
+from service import Sql_contactsOperator
+from service.Sql_complexOrder import *
+import datetime,time
+import pandas as pd
+from datetime import timedelta
+
+class __INIT__(BF):
+    def __init__(self):
+        time_now = datetime.datetime.strptime(datetime.datetime.strftime(datetime.datetime.now(),'%Y-%m-%d %H:%M:%S'),'%Y-%m-%d %H:%M:%S')
+        BF.df_loan_all = getSql_loanAll_v5(str(tuple([self.user_id])).replace(',)',')'))
+        BF._apply_at = time_now
+        BF.df_apply_order = pd.DataFrame()
+        BF.df_loan_rollBack = pd.DataFrame()
+        BF._df_apply = Sql_contactsOperator.getSql_userApplyQuotaRecord(str(tuple([self.user_id])).replace(',)',')'))
+        if BF._df_apply.empty == False:
+            BF._df_apply.sort_values('apply_time',inplace=True)
+            BF._df_apply.reset_index(drop=True,inplace=True)
+            if self.orderId:
+                BF.df_apply_order = BF._df_apply.loc[BF._df_apply['order_no'] == self.orderId]
+                if BF.df_apply_order.empty == False:
+                    BF._apply_at = datetime.datetime.strptime(BF.df_apply_order['apply_time'].astype(str).values[0],'%Y-%m-%d %H:%M:%S')
+
+            if BF.df_loan_all.empty == False:
+                BF.df_loan_all.sort_values('apply_time',inplace=True)
+                BF.df_loan_all.reset_index(drop=True,inplace=True)
+                BF.df_loan_rollBack = get_Loan_rollBack_v5(BF.df_loan_all,BF._apply_at,passdueDay_Negative=True)
+
+
+def get_loanSettleFrom(df_loan_rollBack):
+    """上几笔的订单"""
+    df_loan_result = pd.DataFrame()
+    if df_loan_rollBack.empty == False:
+        # df_loan_rollBack = df_loan_rollBack[df_loan_rollBack["progress"] == 16]
+        # if df_loan_rollBack.empty == False:
+        df_loan_result = df_loan_rollBack[df_loan_rollBack['loan_id'] == df_loan_rollBack['loan_id'].max()]
+    return df_loan_result
+
+
+
+class 历史分期中最大逾期天数可正可负():
+    """delq_days_max_days_v3"""
+    def feature(self):
+        _历史分期中最大逾期天数可正可负 = -9999999
+        df_loan_result = self.df_loan_rollBack
+        if df_loan_result.empty == False:
+            _result =df_loan_result['passdue_day'].max()
+            _历史分期中最大逾期天数可正可负 = _result
+        self._result["历史分期中最大逾期天数可正可负"]["value"] = _历史分期中最大逾期天数可正可负
+        return self._result
+
+
+class 订单的申请次数():
+    """history_apply_cnt_v5"""
+    def feature(self):
+        _订单的申请次数 = -9999999
+        get_new_loan_v4_apply = new_loan_v4_apply(self.user_id,self._apply_at)['count'].values[0]
+        get_xiaoniu_loan_v4_apply = xiaoniu_loan_v4_apply(self.user_id,self._apply_at)['count'].values[0]
+        get_old_loan_v4_apply = old_loan_v4_apply(self.user_id)['count'].values[0]
+        _result = get_new_loan_v4_apply - get_xiaoniu_loan_v4_apply + get_old_loan_v4_apply
+        _订单的申请次数 = _result
+        self._result["订单的申请次数"]["value"] = _订单的申请次数
+        return self._result
+
+
+class 逾期金额比率():
+    """overdue_amount_ratio_v2"""
+    def feature(self):
+        _逾期金额比率 = -9999999
+        df_loan_result = self.df_loan_rollBack
+        if df_loan_result.empty == False:
+            df_loan_result_all = df_loan_result['monthly_repayment'].sum()
+            df_loan_result_passdue_day =df_loan_result[df_loan_result['passdue_day'] >0]['monthly_repayment'].sum()
+            _result = round(df_loan_result_passdue_day / df_loan_result_all,2)
+            _逾期金额比率 = _result
+        self._result["逾期金额比率"]["value"] = _逾期金额比率
+        return self._result
+
+
+class 各笔订单最大的不逾期期数比():
+    """overdue_days_dis_cont_ratio_v2#overdueDaysGt0DisCont 会有少许精度问题--防止四舍六入五平分"""
+    def feature(self):
+        _各笔订单最大的不逾期期数比 = -9999999
+        df_loan_result = self.df_loan_rollBack
+        sum_list = []
+        if df_loan_result.empty == False:
+            _passdue_day_list = list(set(df_loan_result[df_loan_result['passdue_day'] <= 0]['loan_id'].tolist()))
+            df_loan_id = df_loan_result[df_loan_result['loan_id'].isin(_passdue_day_list)]
+            for k,v in df_loan_id.groupby(['loan_id']):
+                v['term_count'] = len(v)
+                v['max_term'] = str(v[v['passdue_day'] >0]['term_no'].max()).replace('nan','0')
+                v['result'] =  v['term_count'].values[0] - int(v['max_term'].values[0])
+                sum_list.append(v['result'].values[0])
+            _result = sum(sum_list) / len(df_loan_result)
+
+            _各笔订单最大的不逾期期数比 =round(_result,2)
+        self._result["各笔订单最大的不逾期期数比"]["value"] = _各笔订单最大的不逾期期数比
+        return self._result
+

features/calc_features/id_infoisValid.py

@@ -41,11 +41,3 @@ class 用户身份证是否过期():
         return self._result
 
 
-# class 用户手机号前三位():
-#     def feature(self):
-#         _用户手机号前三位 = -9999999
-#         id_time= int(str(self._apply_at).split()[0].replace("-", ""))
-#         card_expires_time = card_expires(self.uuid, self.orderId)
-#
-#         self._result['用户手机号前三位']['value'] = _用户手机号前三位
-#         return self._result

features/calc_features/same_trade.py

@@ -4,6 +4,7 @@ from service.addresss_book_sql import *
 import datetime
 from service.Sql_cheat import *
 import re
+import pendulum
 class __INIT__(BF):
     def __init__(self):
         time_now = datetime.datetime.strptime(datetime.datetime.strftime(datetime.datetime.now(),'%Y-%m-%d %H:%M:%S'),'%Y-%m-%d %H:%M:%S')
@@ -16,6 +17,10 @@ class __INIT__(BF):
 
 
 
+def time_float(str_time):
+    """转换为时间戳比大小"""
+    time_now = time.mktime(time.strptime(str(str_time), '%Y-%m-%d %H:%M:%S'))
+    return time_now
 
 class 在贷_同业():
     """same_trade_channel_uncompleted_loan_nums   在贷_同业"""
@@ -58,3 +63,48 @@ class 在途申请单数量_同业():
 
 
 
+
+class 最近一次资产分发拒绝有效期非同业():
+    """最近一次资产分发拒绝有效期非同业 """
+    def feature(self):
+        _最近一次资产分发拒绝有效期非同业 = 0
+        get_no_tongye_info = no_tongye_info(self.uuid)['audit_valid_time'].values[0]
+        if get_no_tongye_info:
+            """将非时间类型转成时间类型的字符串"""
+            get_no_tongye_info_str = str(get_no_tongye_info)
+            str_time = pendulum.parse(get_no_tongye_info_str).to_datetime_string()
+            get_audit_valid_time = time_float(str_time)
+            get_apply_time = time_float(self._apply_at)
+            if get_audit_valid_time > get_apply_time:
+                _最近一次资产分发拒绝有效期非同业 = 1
+            elif get_audit_valid_time < get_apply_time:
+                _最近一次资产分发拒绝有效期非同业 = 0
+            else:
+                _最近一次资产分发拒绝有效期非同业 = 0
+        self._result['最近一次资产分发拒绝有效期非同业']['value'] = _最近一次资产分发拒绝有效期非同业
+        return self._result
+
+
+
+class 最近一次资产分发拒绝有效期同业():
+    """最近一次资产分发拒绝有效期同业 """
+    def feature(self):
+        _最近一次资产分发拒绝有效期同业 = -9999999
+        if self.business_type != 0:
+            get_y_tongye_info = y_tongye_info(self.business_type,self.channel_no,self.uuid)['audit_valid_time'].values[0]
+            if get_y_tongye_info:
+                get_y_tongye_info_str = str(get_y_tongye_info)
+                str_time = pendulum.parse(get_y_tongye_info_str).to_datetime_string()
+                get_audit_valid_time = time_float(str_time)
+                get_apply_time = time_float(self._apply_at)
+                if get_audit_valid_time > get_apply_time:
+                    _最近一次资产分发拒绝有效期同业 = 1
+                elif get_audit_valid_time < get_apply_time:
+                    _最近一次资产分发拒绝有效期同业 = 0
+                else:
+                    _最近一次资产分发拒绝有效期同业 = 0
+            else:
+                _最近一次资产分发拒绝有效期同业 = 0
+        self._result['最近一次资产分发拒绝有效期同业']['value'] = _最近一次资产分发拒绝有效期同业
+        return self._result
+

features/calc_features/vcc_id_valid.py

+# -*- coding:utf-8 -*-
+
+from features.Base_Features import BaseFeatures as BF
+from service.addresss_book_sql import *
+import datetime
+from service.Sql_vcc import *
+import json
+import time
+from pandas import DataFrame
+import numpy as np
+
+import datetime
+from datetime import timedelta
+
+
+
+class __INIT__(BF):
+    def __init__(self):
+        time_now = datetime.datetime.strptime(datetime.datetime.strftime(datetime.datetime.now(),'%Y-%m-%d %H:%M:%S'),'%Y-%m-%d %H:%M:%S')
+        BF._apply_at = time_now
+
+        """vcc申请交易的时间"""
+        if self.orderId:
+            _df = get_orderid_trade(self.orderId)
+            if _df.empty == False:
+                BF._apply_at = datetime.datetime.strptime((_df['trade_time'].astype(str).values[0]),'%Y-%m-%d %H:%M:%S')
+
+class 身份证有效期大于30天():
+    """user_ocr_info#isValid   身份证是否过期  需求待修改"""
+    def feature(self):
+        _身份证有效期大于30天 = -9999999
+        id_time = str(self._apply_at)[0:10]+' ' + '00:00:00'
+        start = (datetime.datetime.strptime(id_time, '%Y-%m-%d %H:%M:%S'))
+        card_expires = str(vcc_idcard_expires(self.uuid,self.orderId))
+        year1 = card_expires[0:4]
+        month1 = card_expires[4:6]
+        day1 = card_expires[6:8]
+        end_time_str = str(year1) + '-' + str(month1) + '-' + str(day1) + ' ' + '00:00:00'
+        end_time = datetime.datetime.strptime(end_time_str, '%Y-%m-%d %H:%M:%S')
+        if card_expires != '-9999999':
+            if (end_time - start).days >=30:
+                _身份证有效期大于30天 = 1
+            elif (end_time - start).days < 30:
+                _身份证有效期大于30天 = 0
+        self._result['身份证有效期大于30天']['value'] = _身份证有效期大于30天
+        return self._result
\ No newline at end of file

service/Sql_cheat.py

@@ -267,43 +267,35 @@ def city_dict_df():
     df = pd.DataFrame(city_dict, pd.Index(range(1)))
     return df
 
-def idcard_dict_df():
-    """标准的身份证归属地码表"""
-    idcard_dict = {'北京': 11,
-                 '天津': 12,
-                 '河北': 13,
-                 '山西': 14,
-                 '内蒙': 15,
-                 '辽宁': 21,
-                 '吉林': 22,
-                 '黑龙江': 23,
-                 '上海': 31,
-                 '江苏': 32,
-                 '浙江': 33,
-                 '安徽': 34,
-                 '福建': 35,
-                 '江西': 36,
-                 '山东': 37,
-                 '河南': 41,
-                 '湖北': 42,
-                 '湖南': 43,
-                 '广东': 44,
-                 '广西': 45,
-                 '海南': 46,
-                 '重庆': 50,
-                 '四川': 51,
-                 '贵州': 52,
-                 '云南': 53,
-                 '西藏': 54,
-                 '陕西': 61,
-                 '甘肃': 62,
-                 '青海': 63,
-                 '宁夏': 64,
-                 '新疆': 65,
-                 '台湾': 71,
-                 '香港': 81,
-                 '澳门': 82,
-                 '海外': 91}
 
-    df = pd.DataFrame(idcard_dict, pd.Index(range(1)))
-    return df
+def decive_info(uuid,start_time,end_time):
+    Sql = """select body,log_time from rc_user_behavior.`anti_fraud_log_kafka`  where  distinct_id = '{0}' and log_time > '{1}'  and log_time < '{2}'  order by log_time; """.format(uuid,start_time,end_time)
+    _df = pd.read_sql(Sql, con=con_device_info)
+    return _df
+
+
+
+
+def no_tongye_info(uuid):
+    sql = """select max(audit_valid_time) audit_valid_time from apply_list 
+            where apply_status=5 and assets_result=0 and enable=1
+            and product_type=0
+            and uuid='{0}';""".format(uuid)
+    _df = pd.read_sql(sql, con=con_business_flow)
+    return _df
+
+def y_tongye_info(bussiness_type,channel_id,uuid):
+    sql = """select max(audit_valid_time) audit_valid_time from apply_list 
+            where apply_status=5 and assets_result=0 and enable=1
+            and product_type='{0}'
+            and channel_id='{1}'
+            and uuid='{2}';""".format(bussiness_type,channel_id,uuid)
+    _df = pd.read_sql(sql, con=con_business_flow)
+    return _df
+
+
+def vcc_data_payment(user_id,time):
+    sql = """select bill_date from acs_bill where user_id = '{0}'
+     and bill_date < '{1}' order by bill_date desc limit 1 """.format(user_id,time)
+    _df = pd.read_sql(sql, con=con_vcc_device_info)
+    return _df
\ No newline at end of file

service/Sql_complexOrder.py

@@ -144,7 +144,7 @@ def getSql_loanAll_v5(user_id):
     _sql = """
         select
          IFNULL(aqr.order_no ,'') order_no,loan.business_type
-        ,loan.user_id,loan.id as loan_id,loan.progress,fest.contract_loan_amount,fest.loan_paid_at,fest.monthly_interest_rate
+        ,loan.user_id,loan.id as loan_id,loan.progress,fest.contract_loan_amount,fest.loan_paid_at,fest.monthly_interest_rate,fest.monthly_repayment
         ,case when loan.created_at >='2018-05-08' then aqr.apply_from
             when loan.created_at < '2018-05-08' then loan.created_from
             else null
@@ -188,5 +188,24 @@ def old_apptime_resufe(user_id,applytime):
  if(l.progress in (6, 8, 19), 1, 0) apply_status,created_from apply_from FROM  loan_application_history l
  LEFT JOIN loan_account_ext e  ON e.loan_id = l.id left join loan_submit_info ls on ls.loan_id = l.id 
  WHERE l.user_id = '{0}'  and l.business_type IN (0, 2, 4,6) AND e.loan_id IS NULL And l.created_at < '{1}';""".format(user_id,applytime)
+    _df = pd.read_sql(_sql,con_tuomin_xyqb)
+    return _df
+
+def new_loan_v4_apply(user_id,applytime):
+    """查找用户所有新申请详情"""
+    _sql = """SELECT COUNT(1) count FROM apply_quota_record  WHERE user_id = '{0}' AND apply_time < '{1}';""".format(user_id,applytime)
+    _df = pd.read_sql(_sql,con_tuomin_xyqb)
+    return _df
+
+def xiaoniu_loan_v4_apply(user_id,applytime):
+    """查询小牛计划退货的申请次数"""
+    _sql = """SELECT COUNT(1) count FROM apply_quota_record a LEFT JOIN loan_account_ext e ON e.order_no = a.order_no 
+    LEFT JOIN loan_application_history l ON l.id = e.loan_id WHERE a.user_id = '{0}' AND l.progress = 48 AND apply_time < '{1}';""".format(user_id,applytime)
+    _df = pd.read_sql(_sql,con_tuomin_xyqb)
+    return _df
+
+def old_loan_v4_apply(user_id):
+    """查找用户所有老的申请详情"""
+    _sql = """SELECT COUNT(1) count FROM loan_application_history l LEFT JOIN loan_account_ext e ON e.loan_id = l.id WHERE l.user_id = '{0}' and l.business_type IN (0,2) AND e.loan_id IS NULL""".format(user_id)
     _df = pd.read_sql(_sql,con_tuomin_xyqb)
     return _df
\ No newline at end of file

service/Sql_vcc.py

@@ -719,4 +719,30 @@ def vcc_credit_phone(userid,phone):
      else '-9999999'
      end as as_type from payment_center.bank_card where uid='{0}' and preserve_phone_no='{1}';""".format(userid,phone)
     _df = pd.read_sql(_sql, con_vcc_transfer_pay)
-    return _df
\ No newline at end of file
+    return _df
+
+
+
+#身份证是否过期
+def vcc_idcard_expires(uuid=None,order_no=None):
+    params = {'uuid': str(uuid),"order_id": str(order_no)}
+    _response = None
+    try:
+        _url = "http://operator.liangkebang.com/auth-center/ex/getIdCardInfo.json"
+        _request = requests.post(_url, data=params)
+        if _request.status_code == 200:
+            _response = _request.json()
+    except Exception as e:
+        print(e.args)
+    if _response:
+        if _response['data']['valid_date_end'] == -9999999:
+            return list([])
+        # if _response['data']['features']['user_ocr_info#validDateEnd']['value'] == "?":
+        #     return list([])
+        ll_idcard = _response['data']['valid_date_end']
+
+        if ll_idcard:
+            if (ll_idcard =='长期') or (ll_idcard =='长期有效'):
+                return '20301230'
+            elif isinstance(ll_idcard.replace(".",""),str):
+                return (ll_idcard.replace(".",""))
\ No newline at end of file