We’ve released new trended attributes for our Cash Advance Repayment Score (CARS), which can be used to predict the likelihood of a user repaying a cash advance. We now have 700+ Attributes in our Attributes Store!
Why this matters
Each sprint, we release 5-10 new Attributes that capture unique cashflow patterns and contribute to our models’ increasing accuracy. You can find the entire list of Attributes and definitions in our docs.
As we further develop our scores, we constantly monitor which Attributes prove to be more predictive than others and adjust our models accordingly.
New Attributes
Attributes that we’ve released and have found to be predictive of repayment in our CARS model include information about a user’s:
- 15/30/45 day cash advance repayment rate
- Number of cash advances paid
- Recurring expenditures other than cash advances
- Debt payments
- and more.
The top 10 most important Attributes in our newest model are:
- 15D_CASH_ADVANCE_REPAYMENT_RATE_PAST_90D
- 15D_CASH_ADVANCE_REPAYMENT_RATE_PAST_180D
- 15D_CASH_ADVANCE_REPAYMENT_RATE_PAST_60D
- 15D_CASH_ADVANCE_REPAYMENT_RATE_PAST_30D
- CASH_ADVANCE_PAID_WITHIN_15D_COUNT_PAST_30D
- PAVE_MERCHANT_NAME_ENCODED
- RECURRING_EXPENDITURES_NEXT_30D
- 30D_CASH_ADVANCE_REPAYMENT_RATE_PAST_180D
- 30D_CASH_ADVANCE_REPAYMENT_RATE_PAST_90D
- CASH_ADVANCE_PAID_WITHIN_30D_COUNT_PAST_30D
You can find the entire list of Attributes and definitions in our docs and can read more about the use cases for Attributes in this blog post.
