README.md

metadata

license: cc-by-4.0
language:
  - ru
tags:
  - bank
  - loan
  - time-series
size_categories:
  - 1M<n<10M
pretty_name: Alfa BKI

Dataset Summary

Alfa BKI is a unique high-quality dataset collected from the real data source of credit history bureaus (in Russian "бюро кредитных историй/БКИ"). It contains the history of corresponding credit products and the applicants' default on the loan.

Supported Tasks and Leaderboards

The dataset is supposed to be used for training models for the classical bank task of predicting the default of the applicant.

Dataset Structure

Data Instances

The example of one sample is provided below

{
  'app_id': 0,
  'history':
  [
    [ 0, 1, 18, 9, 2, 3, 16, 10, 11, 3, 3, 0, 2, 11, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 16, 2, 17, 1, 1, 1, 0, 0, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 3, 4, 1, 0, 0 ],
    [ 0, 2, 18, 9, 14, 14, 12, 12, 0, 3, 3, 0, 2, 11, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 16, 2, 17, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 4, 1, 3, 4, 1, 0, 0 ],
    [ 0, 3, 18, 9, 4, 8, 1, 11, 11, 0, 5, 0, 2, 8, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 15, 2, 17, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 4, 1, 2, 3, 1, 1, 1 ],
    [ 0, 4, 4, 1, 9, 12, 16, 7, 12, 2, 3, 0, 2, 4, 6, 16, 5, 4, 8, 0, 1, 1, 1, 1, 16, 2, 17, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 3, 1, 1, 0, 0 ],
    [ 0, 5, 5, 12, 15, 2, 11, 12, 10, 2, 3, 0, 2, 4, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 16, 2, 17, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 4, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 3, 4, 1, 0, 0 ],
    [ 0, 6, 5, 0, 11, 8, 12, 11, 4, 2, 3, 0, 2, 4, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 9, 5, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 3, 4, 3, 3, 3, 4, 1, 2, 3, 1, 0, 1 ],
    [ 0, 7, 3, 9, 1, 2, 12, 14, 15, 5, 3, 0, 2, 3, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 16, 2, 17, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 4, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 3, 4, 1, 0, 0 ],
    [ 0, 8, 2, 9, 2, 3, 12, 14, 15, 5, 3, 0, 2, 13, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 16, 2, 17, 1, 1, 1, 0, 0, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 3, 4, 1, 0, 0 ],
    [ 0, 9, 1, 9, 11, 13, 14, 8, 2, 5, 1, 0, 2, 11, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 1, 2, 17, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 2, 4, 1, 0, 0 ],
    [ 0, 10, 7, 9, 2, 10, 8, 8, 16, 4, 2, 0, 2, 11, 6, 16, 5, 4, 8, 1, 1, 1, 1, 1, 15, 2, 17, 0, 1, 1, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 3, 4, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 3, 3, 4, 1, 2, 4, 1, 0, 0 ]
  ],
  'flag': 0
}

Data Fields

• id: application ID.
• history: an array of transactions where each credit product is represented as a 37-dimensional array, each element of the array represents a corresponding feature from the following list.
  • id: application ID.
  • rn: serial number of the credit product in the credit history.
  • pre_since_opened: days from the date of opening the loan to the date of data collection.
  • pre_since_confirmed: days from the date of confirmation of the loan information to the date of data collection.
  • pre_pterm: planned number of days from the opening date of the loan to the closing date.
  • pre_fterm: actual number of days from the opening date of the loan to the closing date.
  • pre_till_pclose: planned number of days from the date of data collection to the closing date of the loan.
  • pre_till_fclose: actual number of days from the date of data collection to the closing date of the loan.
  • pre_loans_credit_limit: credit limit.
  • pre_loans_next_pay_summ: amount of the next loan payment.
  • pre_loans_outstanding: remaining unpaid loan amount.
  • pre_loans_total_overdue: current overdue debt.
  • pre_loans_max_overdue_sum: maximum overdue debt.
  • pre_loans_credit_cost_rate: full cost of the loan.
  • pre_loans5: number of delays up to 5 days.
  • pre_loans530: number of delays from 5 to 30 days.
  • pre_loans3060: number of delays from 30 to 60 days.
  • pre_loans6090: number of delays from 60 to 90 days.
  • pre_loans90: the number of delays of more than 90 days.
  • is_zero_loans_5: flag: no delays up to 5 days.
  • is_zero_loans_530: flag: no delays from 5 to 30 days.
  • is_zero_loans_3060: flag: no delays from 30 to 60 days.
  • is_zero_loans_6090: flag: no delays from 60 to 90 days.
  • is_zero_loans90: flag: no delays for more than 90 days.
  • pre_util: ratio of the remaining unpaid loan amount to the credit limit.
  • pre_over2limit: ratio of current overdue debt to the credit limit.
  • pre_maxover2limit: ratio of the maximum overdue debt to the credit limit.
  • is_zero_util: flag: the ratio of the remaining unpaid loan amount to the credit limit is 0.
  • is_zero_over2limit: flag: the ratio of the current overdue debt to the credit limit is 0.
  • is_zero_maxover2limit: flag: the ratio of the maximum overdue debt to the credit limit is 0.
  • enc_paym_{0..n}: monthly payment statuses for the last n months.
  • enc_loans_account_holder_type: type of attitude to credit.
  • enc_loans_credit_status: loan status.
  • enc_loans_account_cur: loan currency.
  • enc_loans_credit_type: type of loan.
  • pclose_flag: flag: the planned number of days from the opening date of the loan to the closing date is not defined.
  • fclose_flag: flag: the actual number of days from the opening date of the loan to the closing date is not determined.
• flag: target, 1 – the fact that the client has defaulted.