提交dev 分支

J_feature/v7模型_延伸特征.ipynb

@@ -836,7 +836,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.6"
+   "version": "3.6.0"
   }
  },
  "nbformat": 4,

J_feature/数据一致性验证.ipynb

@@ -212,7 +212,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.0"
+   "version": "3.6.6"
   }
  },
  "nbformat": 4,

config/concat_sql.py

@@ -80,3 +80,6 @@ con_app_applied_type = create_engine('mysql+pymysql://b_events_w:q3ReZoL9x5asCxc
 """催收二期"""
 con_collection = create_engine('mysql+pymysql://rc-meta-user-pre:YqnqVfIwZFMlkGya@10.17.115.6:4010/rc_meta_data_pre?charset=utf8',echo=False)
 
+"""business flow 线上库"""
+con_business_flow =create_engine('mysql+pymysql://real_data_test:qpRBSarm$Tv*YO!n@172.30.220.16:33314/business_flow?charset=utf8'
+                               ,echo=False)
\ No newline at end of file

features/calc_features/Credit_process.py

+# -*- coding:utf-8 -*-
+
+# -*- coding:utf-8 -*-
+
+from features.Base_Features import BaseFeatures as BF
+from service.addresss_book_sql import *
+import datetime
+from service.Sql_cheat import *
+import re
+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
+        if self.orderId:
+            _df = getSql_Apply_quota_record(self.user_id,self.orderId)
+            if _df.empty == False:
+                BF._apply_at = datetime.datetime.strptime(_df['apply_time'].astype(str).values[0],'%Y-%m-%d %H:%M:%S')
+
+
+class 存在未结清订单数量():
+    """uncompleted_loan_nums 存在未结清订单数量（还款中，在贷）"""
+    def feature(self):
+        _存在未结清订单数量 = -9999999
+        get_quantity = quantity_ofopen_orders(self.user_id,self._apply_at)
+        if get_quantity.empty == False:
+            get_quantity_count = get_quantity['count'].values[0]
+            _存在未结清订单数量 = get_quantity_count
+        self._result['存在未结清订单数量']['value'] = _存在未结清订单数量
+        return self._result
+
+
+class 填写中申请单数量():
+    """user_fill_order_nums  enable = 1填写中申请单数量"""
+    def feature(self):
+        _填写中申请单数量 = -9999999
+        get_quantity = application_forms_nums(self.uuid,self._apply_at)
+        if get_quantity.empty == False:
+            get_quantity_count = get_quantity['count'].values[0]
+            _填写中申请单数量 = get_quantity_count
+        self._result['填写中申请单数量']['value'] = _填写中申请单数量
+        return self._result
+
+
+
+class 用户申请订单在审核中的数量():
+    """new_xyqb_apply_in_credit_v2   用户申请订单在审核中的数量"""
+    def feature(self):
+        _用户申请订单在审核中的数量 = -9999999
+        get_user_order_approval = user_order_approval(self.uuid,self._apply_at)
+        if get_user_order_approval.empty == False:
+            get_user_order_count = get_user_order_approval['count'].values[0]
+            _用户申请订单在审核中的数量 = get_user_order_count
+        self._result['用户申请订单在审核中的数量']['value'] = _用户申请订单在审核中的数量
+        return self._result
+
+
+
+
+# class 在途申请单数量():
+#这个是老的特征 下面的特征进行了更改
+#     """apply_in_verify_num_v2   在途申请单数量（提交申请->完成放款）"""
+#     def feature(self):
+#         _在途申请单数量 = -9999999
+#         get_requisition_quantity_xyqb = requisition_quantity_xyqb(self.user_id,self._apply_at)
+#         get_requisition_quantity_business_flow = requisition_quantity_business_flow(self.uuid,self._apply_at)
+#         if get_requisition_quantity_xyqb.empty == False:
+#             get_quantity_count = get_requisition_quantity_xyqb['count_xyqb'].values[0]
+#         if get_requisition_quantity_business_flow.empty == False:
+#             get_business_flow_count = get_requisition_quantity_business_flow['count_business'].values[0]
+#         _在途申请单数量 = get_quantity_count + get_business_flow_count
+#         self._result['在途申请单数量']['value'] = _在途申请单数量
+#         return self._result
+
+
+
+class 在途申请单数量():
+    """apply_in_verify_num_v2    enable=1 在途申请单数量（提交申请->完成放款）"""
+    def feature(self):
+        _在途申请单数量 = -9999999
+        get_requisition_quantity_xyqb = requisition_quantity_xyqb_new(self.user_id,self._apply_at)
+        get_requisition_quantity_business_flow = requisition_quantity_business_flow_new(self.uuid,self._apply_at)
+        if get_requisition_quantity_xyqb.empty == False:
+            get_quantity_count = get_requisition_quantity_xyqb['count_xyqb'].values[0]
+        if get_requisition_quantity_business_flow.empty == False:
+            get_business_flow_count = get_requisition_quantity_business_flow['count_business'].values[0]
+        _在途申请单数量 = get_quantity_count + get_business_flow_count
+        self._result['在途申请单数量']['value'] = _在途申请单数量
+        return self._result
\ No newline at end of file

features/calc_features/id_place_ownership.py

+# -*- coding:utf-8 -*-
+
+from features.Base_Features import BaseFeatures as BF
+from service.addresss_book_sql import *
+import datetime
+from service.Sql_cheat import *
+from service.Sql_contactsOperator import *
+
+import re
+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
+        if self.orderId:
+            _df = getSql_Apply_quota_record(self.user_id,self.orderId)
+            if _df.empty == False:
+                BF._apply_at = datetime.datetime.strptime(_df['apply_time'].astype(str).values[0],'%Y-%m-%d %H:%M:%S')
+
+
+
+
+class 根据手机号查找归属地编号():
+    """ 根据手机号查找归属地编号   lkb_user_phone_province_code"""
+    def feature(self):
+        _根据手机号查找归属地编号 = 125
+        user_phone = getSql_userPhone(str(tuple([str(self.user_id)])).replace(',)',')'))['phone_no'].tolist()
+        """紧急联系人"""
+        phone_info_city =get_phone_info(user_phone[0])[1]
+        if phone_info_city:
+            get_city_dict_df = city_dict_df()
+            """根据对应的列找到对应的值"""
+            if phone_info_city == '内蒙古':
+                phone_info_city = '内蒙'
+            get_city_dict_value = get_city_dict_df[phone_info_city].values[0]
+            _根据手机号查找归属地编号 = get_city_dict_value
+        self._result['根据手机号查找归属地编号']['value'] = _根据手机号查找归属地编号
+        return self._result
+
+
+class 根据身份证号查找归属地编号():
+    """ 根据身份证号查找归属地编号   user_province_code"""
+    def feature(self):
+        _根据身份证号查找归属地编号 = -9999999
+        id_no_ = getSql_id_no(str(self.user_id))["id_no"].values[0]
+        if id_no_:
+            id_no_split_2 = str(id_no_)[0:2]
+            _根据身份证号查找归属地编号 = int(id_no_split_2)
+        self._result['根据身份证号查找归属地编号']['value'] = _根据身份证号查找归属地编号
+        return self._result
\ No newline at end of file

features/calc_features/test_timeWindow.py

@@ -31,16 +31,32 @@ class __INIT__(BF):
         BF.phon_black_all['type'] = BF.phon_black_all['type'].astype(int)
 
 
-class 本人手机号命中的黑名单类型在0_1内的次数():
+# class 本人手机号命中的黑名单类型在0_1内的次数():
+#     def feature(self,*args):
+#         featureName = args[0]
+#         daysWindow = int(re.findall('\d+',args[1][0])[0])
+#         if daysWindow == 0:
+#             _本人手机号命中的黑名单类型在0_1内的次数 = self.phon_black_all.loc[(self.phon_black_all['type'].isin(self.a_tuple))]['id'].count()
+#         else:
+#             day_window_time = self.get_DayWindow(self._apply_at,daysWindow)
+#             _本人手机号命中的黑名单类型在0_1内的次数 = self.phon_black_all.loc[(self.phon_black_all['createdAt']>day_window_time)
+#                                                              & (self.phon_black_all['type'].isin(self.a_tuple))]['id'].count()
+#         self._result[featureName]["value"] = _本人手机号命中的黑名单类型在0_1内的次数
+#
+#         return self._result
+
+
+
+class 本人手机号命中的黑名单():
     def feature(self,*args):
         featureName = args[0]
         daysWindow = int(re.findall('\d+',args[1][0])[0])
         if daysWindow == 0:
-            _本人手机号命中的黑名单类型在0_1内的次数 = self.phon_black_all.loc[(self.phon_black_all['type'].isin(self.a_tuple))]['id'].count()
+            _本人手机号命中的黑名单 = self.phon_black_all.loc[(self.phon_black_all['type'].isin(self.a_tuple))]['id'].count()
         else:
             day_window_time = self.get_DayWindow(self._apply_at,daysWindow)
-            _本人手机号命中的黑名单类型在0_1内的次数 = self.phon_black_all.loc[(self.phon_black_all['createdAt']>day_window_time)
+            _本人手机号命中的黑名单 = self.phon_black_all.loc[(self.phon_black_all['createdAt']>day_window_time)
                                                              & (self.phon_black_all['type'].isin(self.a_tuple))]['id'].count()
-        self._result[featureName]["value"] = _本人手机号命中的黑名单类型在0_1内的次数
+        self._result[featureName]["value"] = _本人手机号命中的黑名单
 
-        return self._result
+        return self._result
\ No newline at end of file

service/Sql_cheat.py

@@ -32,8 +32,8 @@ def get_if_apply_type_sql(uuid=None, order_no=None):
     params = {'user_uuid': str(uuid), 'codes': 'user_loan_type_v4',
               "order_id": str(order_no), "business_type": 0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:
@@ -70,7 +70,169 @@ where user_id ='{0}' and apply_time < '{1}'  ORDER BY apply_time desc limit 1'''
     return _df
 
 
+def quantity_ofopen_orders(userid,apply_time):
+    """存在未结清订单数量（还款中，在贷）"""
+    Sql = '''select count(*)  as count from loan_application_manifest_history a
+left join repayment_plan b on a.loan_application_history_id=b.loan_application_history_id
+and a.contract_term=b.term_no 
+where (b.repayment_status not in (3,4) or (b.repayment_status=3 and b.repaid_at>'{0}'))
+and a.transaction_status in (2,5) and a.loan_paid_at<'{0}'
+and a.user_id='{1}' ;'''.format(apply_time,userid,)
+    _df = pd.read_sql(Sql, con=con_tuomin_xyqb)
+    return _df
+
+
+def application_forms_nums(uuid,apply_time):
+    """填写中申请单数量"""
+    Sql = """select count(*) as count from apply_list
+             where apply_status=1 and uuid='{0}' and enable=1 and created_at< '{1}';""".format(uuid,apply_time)
+    _df = pd.read_sql(Sql, con=con_business_flow)
+    return _df
+
 
+def requisition_quantity_xyqb(userid,apply_time):
+    """xyqb 在在途申请单数量"""
+    Sql = """SELECT count(*) as count_xyqb  FROM assign_funding_record a
+        left join assets_order b on a.asset_order_id=b.id
+        left join waiting_funding_corp_operate_people c on b.loan_id=c.loan_application_history_id
+        WHERE a.assign_status=3 and (c.id is null or c.funding_corp_progress not in (5,6))
+        and a.user_id='{0}' and a.created_at<'{1}';""".format(userid,apply_time)
+    _df = pd.read_sql(Sql, con=con_tuomin_xyqb)
+    return _df
+
+def requisition_quantity_business_flow(uuid,apply_time):
+    """business_flow 在在途申请单数量"""
+    Sql = """SELECT count(*) as count_business FROM apply_list WHERE (apply_status in (2,3) or (apply_status =4 and audit_result<>0))
+              and uuid ='{0}' and created_at< '{1}';""".format(uuid,apply_time)
+    _df = pd.read_sql(Sql, con=con_business_flow)
+    return _df
 
 
+def user_order_approval(uuid,apply_time):
+    Sql = """select count(id) count from apply_list where enable = 1 
+        and apply_status =3 and  uuid ='{0}' and created_at< '{1}'; """.format(uuid,apply_time)
+    _df = pd.read_sql(Sql, con=con_business_flow)
+    return _df
 
+def requisition_quantity_xyqb_new(userid,apply_time):
+    """xyqb 在在途申请单数量"""
+    Sql = """SELECT count(*) count_xyqb FROM assign_funding_record a
+    left join assets_order b on a.asset_order_id=b.id
+    left join waiting_funding_corp_operate_people c on b.loan_id=c.loan_application_history_id
+    WHERE a.assign_status=3 and (c.id is null or c.funding_corp_progress not in (5,6))
+    and a.user_id='{0}' and a.created_at<'{1}';""".format(userid,apply_time)
+    _df = pd.read_sql(Sql, con=con_tuomin_xyqb)
+    return _df
+
+def requisition_quantity_business_flow_new(uuid,apply_time):
+    """business_flow 在在途申请单数量"""
+    Sql = """SELECT count(*) count_business FROM apply_list
+    WHERE (apply_status in (2,3) or (apply_status =4 and audit_result<>0)) and (assets_result<>2 or assets_result is null)
+    and uuid ='{0}' and enable=1 and created_at< '{1}';""".format(uuid,apply_time)
+    _df = pd.read_sql(Sql, con=con_business_flow)
+    return _df
+
+
+
+
+
+
+from phone import Phone
+
+def get_phone_info(phone_num):
+    """根据手机号查询省份"""
+    phone_info = Phone().find(phone_num)
+    try:
+        phone = phone_info['phone']
+        province = phone_info['province']  # 省
+        city = phone_info['city']  # 城市
+        zip_code = phone_info['zip_code']  # 邮编
+        area_code = phone_info['area_code']  # 区号
+        phone_type = phone_info['phone_type']  # 运营商
+    except:
+        print('----- the phone number is empty, please check your phone number! -----')
+    return phone, province, city, zip_code, area_code, phone_type
+
+
+
+
+def city_dict_df():
+    """标准的归属地码表"""
+    city_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(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

service/Sql_contactsOperator.py

@@ -713,8 +713,8 @@ def get_user_bill_info_rows(uuid,_created_at=None,time_from=0):
     """获取手机话费账单"""
     params = {'user_uuid':str(uuid),'codes':'user_bill_info_rows'}
     headers = {
-        'rc_auth_key':'rc_developer',
-        'rc_auth_secret':'rc_secret'
+        'rc_auth_key':'rc_offline',
+        'rc_auth_secret':"9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:
@@ -742,8 +742,8 @@ def get_lkb_user_phone_location(uuid):
 
     params = {'user_uuid':str(uuid),'codes':'lkb_user_phone_location'}
     headers = {
-        'rc_auth_key':'rc_developer',
-        'rc_auth_secret':'rc_secret'
+        'rc_auth_key':'rc_offline',
+        'rc_auth_secret':"9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:

service/Sql_tongdun.py

@@ -17,8 +17,8 @@ def get_td_qu_rules(uuid):
 
     params = {'user_uuid':str(uuid),'codes':'td_qu_rules'}
     headers = {
-        'rc_auth_key':'rc_developer',
-        'rc_auth_secret':'rc_secret'
+        'rc_auth_key':'rc_offline',
+        'rc_auth_secret':"9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:

service/Sql_vcc.py

@@ -151,8 +151,8 @@ def get_lxr_list_hmd(uuid=None,order_no=None):
     params = {'user_uuid':str(uuid) , 'codes': 'user_contact_book_info#d30_contact_nums',
               "order_id":str(order_no),"business_type":0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     _df = pd.DataFrame()
@@ -238,8 +238,8 @@ def get30day_addressbook(uuid=None,order_no=None):
     params = {'user_uuid':str(uuid) , 'codes': 'user_contact_book_info#d30_contact_nums',
               "order_id":str(order_no),"business_type":0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     _df = pd.DataFrame()
@@ -343,8 +343,8 @@ def Nation(uuid=None,order_no=None):
     params = {'user_uuid': str(uuid), 'codes': 'user_ocr_info#nation',
               "order_id": str(order_no), "business_type": 0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:
@@ -456,8 +456,8 @@ def city_sign(uuid=None,order_no=None):
     params = {'user_uuid':str(uuid) , 'codes': 'user_ocr_info#address',
               "order_id":str(order_no),"business_type":0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:
@@ -518,8 +518,8 @@ def card_expires(uuid=None,order_no=None):
     params = {'user_uuid': str(uuid), 'codes': 'user_ocr_info#validDateEnd',
               "order_id": str(order_no), "business_type": 0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:

service/addresss_book_sql.py

@@ -179,8 +179,8 @@ def get_lxr_list(uuid=None,order_no=None):
     params = {'user_uuid':str(uuid) , 'codes': 'user_contact_book_info#d7_contact_nums',
               "order_id":str(order_no),"business_type":0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     try:
@@ -209,8 +209,8 @@ def get_lxr_list_hmd(uuid=None,order_no=None):
     params = {'user_uuid':str(uuid) , 'codes': 'user_contact_book_info#d7_contact_nums',
               "order_id":str(order_no),"business_type":0}
     headers = {
-        'rc_auth_key': 'rc_developer',
-        'rc_auth_secret': 'rc_secret'
+        'rc_auth_key': 'rc_offline',
+        'rc_auth_secret': "9d9}bc24!e1z1x3`(x~4r29d$+45n3)'zb696b$85e>_]p2&4f{,a3~8b3e_ldt^"
     }
     _response = None
     _df = pd.DataFrame()