某银行的个人征信数据

某银行的公开数据
可用的训练数据包括用户的基本属性user_info.txt、银行流水记录bank_detail.txt、用户浏览行为browse_history.txt、信用卡账单记录bill_detail.txt、放款时间loan_time.txt，以及这些顾客是否发生逾期行为的记录overdue.txt。（注意：并非每一位用户都有非常完整的记录，如有些用户并没有信用卡账单记录，有些用户却没有银行流水记录。） 相应地，还有用于测试的用户的基本属性、银行流水、信用卡账单记录、浏览行为、放款时间等数据信息，以及待预测用户的id列表。 脱敏处理：(a) 隐藏了用户的id信息；(b) 将用户属性信息全部数字化；(c) 将时间戳和所有金额的值都做了函数变换。
（1）用户的基本属性user_info.txt。共6个字段，其中字段性别为0表示性别未知。 用户id,性别,职业,教育程度,婚姻状态,户口类型 6346,1,2,4,4,2 2583,2,2,2,2,1 9530,1,2,4,4,2 6707,0,2,3,3,2 （2）银行流水记录bank_detail.txt。共5个字段，其中，第2个字段，时间戳为0表示时间未知；第3个字段，交易类型有两个值，1表示支出、0表示收入；第5个字段，工资收入标记为1时，表示工资收入。 用户id,时间戳,交易类型,交易金额,工资收入标记 6951,5894316387,0,13.756664,0 6951,5894321388,1,13.756664,0 18418,5896951231,1,11.978812,0 18418,5897181971,1,12.751543,0 18418,5897293906,0,14.456463,1 （3）用户浏览行为browse_history.txt。共4个字段。其中，第2个字段，时间戳为0表示时间未知。 用户id,时间戳,浏览行为数据,浏览子行为编号 34724,5926003545,172,1 34724,5926003545,163,4 34724,5926003545,38,7 67215,5932800403,163,4 67215,5932800403,138,4 67215,5932800403,109,7 （4）信用卡账单记录bill_detail.txt。共15个字段，其中，第2个字段，时间戳为0表示时间未知。为方便浏览，字段以表格的形式给出。

文件示例如下： 用户id,账单时间戳,银行id,上期账单金额,上期还款金额,信用卡额度,本期账单余额,本期账单最低还款额,消费笔数,本期账单金额,调整金额,循环利息,可用金额,预借现金额度,还款状态 3147,5906744363,6,18.626118,18.661937,20.664418,18.905766,17.847133,1,0.000000,0.000000,0.000000,0.000000,19.971271,0 22717,5934018585,3,0.000000,0.000000,20.233635,18.574069,18.396785,0,0.000000,0.000000,0.000000,0.000000,0.000000,0 （5）放款时间信息loan_time.txt。共2个字段，用户id和放款时间。 用户id,放款时间 1,5914855887 2,5914855887 3,5914855887 （6）顾客是否发生逾期行为的记录overdue.txt。共2个字段。样本标签为1，表示逾期30天以上；样本标签为0，表示逾期10天以内。注意：逾期10天~30天之内的用户，并不在此问题考虑的范围内。用于测试的用户，只提供id列表，文件名为testUsers.csv。 用户id,样本标签 1,1 2,0 3,1

This is a public dataset from a certain bank. The available training data includes user basic attributes (stored in user_info.txt), bank transaction records (bank_detail.txt), user browsing behavior logs (browse_history.txt), credit card bill records (bill_detail.txt), loan disbursement time data (loan_time.txt), and records of whether customers have overdue payments (overdue.txt). (Note: Not every user has a complete set of records. For example, some users do not have credit card bill records, while others lack bank transaction records.)

Correspondingly, there are also test data including user basic attributes, bank transaction records, credit card bill records, browsing behavior logs, loan disbursement time data, etc., as well as the ID list of users to be predicted.

Data anonymization measures: (a) User ID information is hidden; (b) All user attribute information is digitized; (c) Timestamps and all monetary values are transformed via mathematical functions.

(1) User basic attributes (user_info.txt): This file contains 6 fields. A gender value of 0 denotes unknown gender.
The schema is: User ID, Gender, Occupation, Education Level, Marital Status, Household Registration Type
Sample rows:
6346,1,2,4,4,2
2583,2,2,2,2,1
9530,1,2,4,4,2
6707,0,2,3,3,2

(2) Bank transaction records (bank_detail.txt): This file contains 5 fields. A timestamp value of 0 indicates unknown time. The transaction type has two possible values: 1 represents expenditure, 0 represents income. A salary income flag of 1 indicates that the transaction is salary income.
The schema is: User ID, Timestamp, Transaction Type, Transaction Amount, Salary Income Flag
Sample rows:
6951,5894316387,0,13.756664,0
6951,5894321388,1,13.756664,0
18418,5896951231,1,11.978812,0
18418,5897181971,1,12.751543,0
18418,5897293906,0,14.456463,1

(3) User browsing behavior logs (browse_history.txt): This file contains 4 fields. A timestamp value of 0 indicates unknown time.
The schema is: User ID, Timestamp, Browsing Behavior Data, Browsing Sub-behavior Number
Sample rows:
34724,5926003545,172,1
34724,5926003545,163,4
34724,5926003545,38,7
67215,5932800403,163,4
67215,5932800403,138,4
67215,5932800403,109,7

(4) Credit card bill records (bill_detail.txt): This file contains 15 fields. A timestamp value of 0 indicates unknown time. The fields are presented in tabular form for clarity.
The schema is: User ID, Bill Timestamp, Bank ID, Previous Bill Amount, Previous Repayment Amount, Credit Card Limit, Current Bill Balance, Minimum Repayment Amount for Current Bill, Number of Transactions, Current Bill Amount, Adjustment Amount, Revolving Interest, Available Amount, Cash Advance Limit, Repayment Status
Sample rows:
3147,5906744363,6,18.626118,18.661937,20.664418,18.905766,17.847133,1,0.000000,0.000000,0.000000,0.000000,19.971271,0
22717,5934018585,3,0.000000,0.000000,20.233635,18.574069,18.396785,0,0.000000,0.000000,0.000000,0.000000,0.000000,0

(5) Loan disbursement time data (loan_time.txt): This file contains 2 fields: User ID and Loan Disbursement Time.
The schema is: User ID, Loan Disbursement Time
Sample rows:
1,5914855887
2,5914855887
3,5914855887

(6) Overdue payment records (overdue.txt): This file contains 2 fields. The sample label 1 indicates that the user has overdue payments for more than 30 days; label 0 indicates that the user has overdue payments within 10 days. Note: Users with overdue periods between 10 and 30 days are not considered in this task. For test users, only their ID lists are provided, stored in testUsers.csv.
The schema is: User ID, Sample Label
Sample rows:
1,1
2,0
3,1

搜集汇总

背景与挑战

背景概述

该数据集为某银行的公开个人征信数据，包含用户基本属性、银行流水、浏览行为、信用卡账单和放款时间等多类文件，用于训练和测试。数据经过脱敏处理，如隐藏用户ID和数字化属性，旨在预测用户是否发生逾期行为。